code
stringlengths
101
5.91M
class StableDiffusionInpaintPipelineLegacy(metaclass=DummyObject): _backends = ['torch', 'transformers'] def __init__(self, *args, **kwargs): requires_backends(self, ['torch', 'transformers']) def from_config(cls, *args, **kwargs): requires_backends(cls, ['torch', 'transformers']) def from_pretrained(cls, *args, **kwargs): requires_backends(cls, ['torch', 'transformers'])
def load_data_table(table, image_dir, corrupt_images=None): print('Loading dataframe for', table) df = pd.read_csv(table) print('Found', len(df), 'images in table') df['filepath'] = df.apply((lambda row: get_image_filepath(row, image_dir)), axis=1) len_before = len(df) if (corrupt_images is not None): df = df[df.apply((lambda row: ('{}/{}'.format(row['PTID'], row['Visit']) not in corrupt_images)), axis=1)] print('Filtered out', (len_before - len(df)), 'of', len_before, 'images because of failed preprocessing') len_before = len(df) df = df[map(os.path.exists, df['filepath'])] print('Filtered out', (len_before - len(df)), 'of', len_before, 'images because of missing files') len_before = len(df) df = df[(df['DX'] != 'MCI')] print('Filtered out', (len_before - len(df)), 'of', len_before, 'images that were MCI') print('Final dataframe contains', len(df), 'images from', len(df['PTID'].unique()), 'patients') print() return df
class TrialOutput(object): def __init__(self, config, model_path): self.config = config self.model_path = model_path
class Entity(object): def __init__(self): self.name = '' self.size = 0.05 self.movable = False self.collide = True self.density = 25.0 self.color = None self.max_speed = None self.accel = None self.max_a_speed = None self.state = EntityState() self.initial_mass = 1.0 def mass(self): return self.initial_mass
class ConvolutionalAutoencoder(torch.nn.Module): def __init__(self): super().__init__() self.conv1 = torch.nn.Conv2d(3, 32, 3, padding=1) self.conv2 = torch.nn.Conv2d(32, 64, 3, padding=1) self.conv3 = torch.nn.Conv2d(64, 128, 3, padding=1) self.conv4 = torch.nn.Conv2d(128, latent_code_size, 3, padding=1) self.trans1 = torch.nn.ConvTranspose2d(latent_code_size, 128, 3, padding=1) self.trans2 = torch.nn.ConvTranspose2d(128, 64, 3, padding=1) self.trans3 = torch.nn.ConvTranspose2d(64, 32, 3, padding=1) self.trans4 = torch.nn.ConvTranspose2d(32, 3, 3, padding=1) self.mp = torch.nn.MaxPool2d(2, return_indices=True) self.up = torch.nn.MaxUnpool2d(2) self.relu = torch.nn.ReLU() def encoder(self, x): x = self.conv1(x) x = self.relu(x) s1 = x.size() (x, ind1) = self.mp(x) x = self.conv2(x) x = self.relu(x) s2 = x.size() (x, ind2) = self.mp(x) x = self.conv3(x) x = self.relu(x) s3 = x.size() (x, ind3) = self.mp(x) x = self.conv4(x) x = self.relu(x) return (x, ind1, s1, ind2, s2, ind3, s3) def decoder(self, x, ind1, s1, ind2, s2, ind3, s3): x = self.trans1(x) x = self.relu(x) x = self.up(x, ind3, output_size=s3) x = self.trans2(x) x = self.relu(x) x = self.up(x, ind2, output_size=s2) x = self.trans3(x) x = self.relu(x) x = self.up(x, ind1, output_size=s1) x = self.trans4(x) x = self.relu(x) return x def forward(self, x): (x, ind1, s1, ind2, s2, ind3, s3) = self.encoder(x) output = self.decoder(x, ind1, s1, ind2, s2, ind3, s3) return output
class CiteseerBiGraph(BaseData): def __init__(self, data_root: Optional[str]=None) -> None: super().__init__('citeseer_bigraph', data_root) self._content = {'num_u_classes': 6, 'num_u_vertices': 1237, 'num_v_vertices': 742, 'num_edges': 1665, 'dim_u_features': 3703, 'dim_v_features': 3703, 'u_features': {'upon': [{'filename': 'u_features.pkl', 'md5': 'd8c1ccd6026cbb1f05cc3c534b239e00'}], 'loader': load_from_pickle, 'preprocess': [to_tensor, partial(norm_ft, ord=1)]}, 'v_features': {'upon': [{'filename': 'v_features.pkl', 'md5': '7ca1d16ad557945f9b66ef6ac40c0210'}], 'loader': load_from_pickle, 'preprocess': [to_tensor, partial(norm_ft, ord=1)]}, 'edge_list': {'upon': [{'filename': 'edge_list.pkl', 'md5': '2a632085fb8f691af6399fbb71dc1f67'}], 'loader': load_from_pickle}, 'u_labels': {'upon': [{'filename': 'u_labels.pkl', 'md5': 'b4d0034c29f6f5b6da17f3037c2af605'}], 'loader': load_from_pickle, 'preprocess': [to_long_tensor]}}
def make_chem_data(rule, train=True): is_train = ('train' if train else 'test') x_data = np.load(osp.join(foldername, (((('logic_' + str(rule)) + '_X_') + is_train) + '.npy'))) y_data = np.load(osp.join(foldername, (((('logic_' + str(rule)) + '_Y_') + is_train) + '.npy'))) return (x_data, y_data)
def test_pvt(): with pytest.raises(TypeError): PyramidVisionTransformer(pretrained=123) with pytest.raises(AssertionError): PyramidVisionTransformer(pretrain_img_size=(224, 224, 224)) temp = torch.randn((1, 3, 224, 224)) model = PyramidVisionTransformer(pretrain_img_size=224, use_abs_pos_embed=True) model.init_weights() model(temp) temp = torch.randn((1, 3, 32, 32)) model = PyramidVisionTransformer() outs = model(temp) assert (outs[0].shape == (1, 64, 8, 8)) assert (outs[1].shape == (1, 128, 4, 4)) assert (outs[2].shape == (1, 320, 2, 2)) assert (outs[3].shape == (1, 512, 1, 1)) temp = torch.randn((1, 3, 33, 33)) model = PyramidVisionTransformer() outs = model(temp) assert (outs[0].shape == (1, 64, 8, 8)) assert (outs[1].shape == (1, 128, 4, 4)) assert (outs[2].shape == (1, 320, 2, 2)) assert (outs[3].shape == (1, 512, 1, 1)) temp = torch.randn((1, 3, 112, 137)) model = PyramidVisionTransformer() outs = model(temp) assert (outs[0].shape == (1, 64, 28, 34)) assert (outs[1].shape == (1, 128, 14, 17)) assert (outs[2].shape == (1, 320, 7, 8)) assert (outs[3].shape == (1, 512, 3, 4))
def seresnet200b(**kwargs): return get_seresnet(blocks=200, conv1_stride=False, model_name='seresnet200b', **kwargs)
.parametrize('model_name, with_cls_token, share_embeddings, return_dataframe', [('saint', False, True, False), ('saint', True, True, False), ('saint', False, False, False), ('saint', False, True, True), ('saint', True, True, True), ('saint', False, False, True), ('fttransformer', False, True, False), ('fttransformer', True, True, False), ('fttransformer', False, False, False), ('fttransformer', False, True, True), ('fttransformer', True, True, True), ('fttransformer', False, False, True), ('tabfastformer', False, True, False), ('tabfastformer', True, True, False), ('tabfastformer', False, False, False), ('tabfastformer', False, True, True), ('tabfastformer', True, True, True), ('tabfastformer', False, False, True), ('tabperceiver', False, True, False), ('tabperceiver', False, False, False), ('tabperceiver', False, True, False), ('tabperceiver', False, False, True)]) def test_transformer_family_models(model_name, with_cls_token, share_embeddings, return_dataframe): embed_cols = ['a', 'b'] cont_cols = ['c', 'd'] tab_preprocessor = TabPreprocessor(cat_embed_cols=embed_cols, continuous_cols=cont_cols, for_transformer=True, with_cls_token=with_cls_token, shared_embed=share_embeddings) X_tab = tab_preprocessor.fit_transform(df_init) params = {'column_idx': tab_preprocessor.column_idx, 'cat_embed_input': tab_preprocessor.cat_embed_input, 'continuous_cols': tab_preprocessor.continuous_cols} deeptabular = _build_model(model_name, params) model = WideDeep(deeptabular=deeptabular) t2v = Tab2Vec(model, tab_preprocessor, return_dataframe=return_dataframe) t2v_out = t2v.transform(df_t2v) out_dim = ((len(embed_cols) + len(cont_cols)) * deeptabular.input_dim) assert (t2v_out.shape[1] == out_dim)
class TFAlbertModelTester(): def __init__(self, parent, batch_size=13, seq_length=7, is_training=True, use_input_mask=True, use_token_type_ids=True, use_labels=True, vocab_size=99, embedding_size=16, hidden_size=32, num_hidden_layers=5, num_attention_heads=4, intermediate_size=37, hidden_act='gelu', hidden_dropout_prob=0.1, attention_probs_dropout_prob=0.1, max_position_embeddings=512, type_vocab_size=16, type_sequence_label_size=2, initializer_range=0.02, num_labels=3, num_choices=4, scope=None): self.parent = parent self.batch_size = 13 self.seq_length = 7 self.is_training = True self.use_input_mask = True self.use_token_type_ids = True self.use_labels = True self.vocab_size = 99 self.embedding_size = 16 self.hidden_size = 32 self.num_hidden_layers = 5 self.num_attention_heads = 4 self.intermediate_size = 37 self.hidden_act = 'gelu' self.hidden_dropout_prob = 0.1 self.attention_probs_dropout_prob = 0.1 self.max_position_embeddings = 512 self.type_vocab_size = 16 self.type_sequence_label_size = 2 self.initializer_range = 0.02 self.num_labels = 3 self.num_choices = 4 self.scope = None def prepare_config_and_inputs(self): input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size) input_mask = None if self.use_input_mask: input_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2) token_type_ids = None if self.use_token_type_ids: token_type_ids = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size) sequence_labels = None token_labels = None choice_labels = None if self.use_labels: sequence_labels = ids_tensor([self.batch_size], self.type_sequence_label_size) token_labels = ids_tensor([self.batch_size, self.seq_length], self.num_labels) choice_labels = ids_tensor([self.batch_size], self.num_choices) config = AlbertConfig(vocab_size=self.vocab_size, hidden_size=self.hidden_size, embedding_size=self.embedding_size, num_hidden_layers=self.num_hidden_layers, num_attention_heads=self.num_attention_heads, intermediate_size=self.intermediate_size, hidden_act=self.hidden_act, hidden_dropout_prob=self.hidden_dropout_prob, attention_probs_dropout_prob=self.attention_probs_dropout_prob, max_position_embeddings=self.max_position_embeddings, type_vocab_size=self.type_vocab_size, initializer_range=self.initializer_range) return (config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels) def create_and_check_albert_model(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels): model = TFAlbertModel(config=config) inputs = {'input_ids': input_ids, 'attention_mask': input_mask, 'token_type_ids': token_type_ids} result = model(inputs) inputs = [input_ids, input_mask] result = model(inputs) result = model(input_ids) self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size)) self.parent.assertEqual(result.pooler_output.shape, (self.batch_size, self.hidden_size)) def create_and_check_albert_for_pretraining(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels): config.num_labels = self.num_labels model = TFAlbertForPreTraining(config=config) inputs = {'input_ids': input_ids, 'attention_mask': input_mask, 'token_type_ids': token_type_ids} result = model(inputs) self.parent.assertEqual(result.prediction_logits.shape, (self.batch_size, self.seq_length, self.vocab_size)) self.parent.assertEqual(result.sop_logits.shape, (self.batch_size, self.num_labels)) def create_and_check_albert_for_masked_lm(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels): model = TFAlbertForMaskedLM(config=config) inputs = {'input_ids': input_ids, 'attention_mask': input_mask, 'token_type_ids': token_type_ids} result = model(inputs) self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.vocab_size)) def create_and_check_albert_for_sequence_classification(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels): config.num_labels = self.num_labels model = TFAlbertForSequenceClassification(config=config) inputs = {'input_ids': input_ids, 'attention_mask': input_mask, 'token_type_ids': token_type_ids} result = model(inputs) self.parent.assertEqual(result.logits.shape, (self.batch_size, self.num_labels)) def create_and_check_albert_for_question_answering(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels): model = TFAlbertForQuestionAnswering(config=config) inputs = {'input_ids': input_ids, 'attention_mask': input_mask, 'token_type_ids': token_type_ids} result = model(inputs) self.parent.assertEqual(result.start_logits.shape, (self.batch_size, self.seq_length)) self.parent.assertEqual(result.end_logits.shape, (self.batch_size, self.seq_length)) def create_and_check_albert_for_multiple_choice(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels): config.num_choices = self.num_choices model = TFAlbertForMultipleChoice(config=config) multiple_choice_inputs_ids = tf.tile(tf.expand_dims(input_ids, 1), (1, self.num_choices, 1)) multiple_choice_input_mask = tf.tile(tf.expand_dims(input_mask, 1), (1, self.num_choices, 1)) multiple_choice_token_type_ids = tf.tile(tf.expand_dims(token_type_ids, 1), (1, self.num_choices, 1)) inputs = {'input_ids': multiple_choice_inputs_ids, 'attention_mask': multiple_choice_input_mask, 'token_type_ids': multiple_choice_token_type_ids} result = model(inputs) self.parent.assertListEqual(list(result['logits'].shape), [self.batch_size, self.num_choices]) def create_and_check_albert_for_token_classification(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels): config.num_labels = self.num_labels model = TFAlbertForTokenClassification(config=config) inputs = {'input_ids': input_ids, 'attention_mask': input_mask, 'token_type_ids': token_type_ids} result = model(inputs) self.parent.assertListEqual(list(result['logits'].shape), [self.batch_size, self.seq_length, self.num_labels]) def prepare_config_and_inputs_for_common(self): config_and_inputs = self.prepare_config_and_inputs() (config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels) = config_and_inputs inputs_dict = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': input_mask} return (config, inputs_dict)
def _pcfg(url='', hf_hub='', mean=None, std=None): return dict(url=url, hf_hub=hf_hub, mean=mean, std=std)
def autoselect(method: str, source: Optional[str]=None, backend: Optional[str]=None, **kwargs) -> StainNormalizer: if (backend is None): backend = sf.backend() if (backend == 'tensorflow'): import slideflow.norm.tensorflow BackendNormalizer = sf.norm.tensorflow.TensorflowStainNormalizer elif (backend == 'torch'): import slideflow.norm.torch BackendNormalizer = sf.norm.torch.TorchStainNormalizer elif (backend == 'opencv'): BackendNormalizer = StainNormalizer else: raise errors.UnrecognizedBackendError if (method in BackendNormalizer.normalizers): normalizer = BackendNormalizer(method, **kwargs) else: normalizer = StainNormalizer(method, **kwargs) if ((source is not None) and (source != 'dataset')): normalizer.fit(source) return normalizer
def get_hashes_and_lines(raw_line): hash = hashlib.md5(raw_line).hexdigest() return (hash, raw_line)
def plot_precision_recall(data, ax=None): df = pandas.DataFrame({'threshold': numpy.linspace(0, 1.0, 50, endpoint=False)}) micro = df.apply((lambda r: score(data, average='micro', threshold=r.threshold)), axis=1) micro['threshold'] = df.threshold micro['micro'] = micro.precision macro = df.apply((lambda r: score(data, average='macro', threshold=r.threshold)), axis=1) macro['threshold'] = df.threshold macro['macro'] = macro.precision (fig, (ax2, ax)) = plt.subplots(1, 2, figsize=(8, 4)) macro.plot.line(ax=ax2, y=['precision', 'recall'], x='threshold', ylim=(0.0, 1.0)) micro.plot.line(ax=ax, y='micro', x='recall', ylim=(0.0, 1.0), xlim=(0, 1)) macro.plot.line(ax=ax, y='macro', x='recall', ylim=(0.0, 1.0), xlim=(0, 1)) ax.set_yticks(numpy.arange(0.0, 1.0, 0.1)) ax.grid(True) ax.set_ylabel('Precision') ax.set_xlabel('Recall') ax.set_aspect('equal') ax2.set_aspect('equal') ax2.set_yticks(numpy.arange(0.0, 1.0, 0.1)) ax2.grid(True) ax2.set_xlabel('Probability threshold for "unknown" class') ax2.set_ylabel('Performance metric') fig.tight_layout()
def hawq_top(fp32_model, q_model, dataloader, criterion, enable_act): orig_eval = True if fp32_model.training: orig_eval = False fp32_model.eval() ht = HessianTrace(fp32_model, dataloader=dataloader, q_model=q_model) traces = ht.get_avg_traces(enable_act, num_sample=0) op_to_traces = traces['weight'] q_model_state_dict = {} for key in q_model.state_dict().keys(): length = len('_model.') new_key = key[length:] q_model_state_dict[new_key] = q_model.state_dict()[key] weight_quant_loss = compare_weights(ht.model.state_dict(), q_model_state_dict) pertur_lst = {} for key in weight_quant_loss: op_float_tensor = weight_quant_loss[key]['float'] op_qnt_tensor = weight_quant_loss[key]['quantized'].dequantize() diff_l2 = (torch.norm((op_float_tensor - op_qnt_tensor), p=2) ** 2) pertur_lst[key] = diff_l2 if enable_act: act_to_traces = traces['activation'] for (trace_i, pertur_i, act_i) in zip(op_to_traces.keys(), pertur_lst.keys(), act_to_traces.keys()): op_to_traces[trace_i] = ((pertur_lst[pertur_i] * op_to_traces[trace_i]) + act_to_traces[act_i]) else: for (trace_i, pertur_i) in zip(op_to_traces.keys(), pertur_lst.keys()): op_to_traces[trace_i] = op_to_traces[trace_i] if (orig_eval is False): fp32_model.train() return op_to_traces
def get_random_walk_eval(sos_key, nodes, edges, nsample=5): group = [] cnt = 0 while (cnt <= nsample): rand_1 = random_walk_from_sos(sos_key, nodes, edges) sample_sent1 = tokenizer.decode(rand_1, skip_special_tokens=True) group.append(sample_sent1) cnt += 1 stat = {} b = self_bleu(group) ed_dis = self_edit_distance(group) stat['rand_walk_self_bleu'] = b stat['rand_walk_self_edit'] = ed_dis return stat
def _conv_flops_compute(input, weight, bias=None, stride=1, padding=0, dilation=1, groups=1): assert ((weight.shape[1] * groups) == input.shape[1]) batch_size = input.shape[0] in_channels = input.shape[1] out_channels = weight.shape[0] kernel_dims = list(weight.shape[2:]) input_dims = list(input.shape[2:]) length = len(input_dims) paddings = (padding if (type(padding) is tuple) else ((padding,) * length)) strides = (stride if (type(stride) is tuple) else ((stride,) * length)) dilations = (dilation if (type(dilation) is tuple) else ((dilation,) * length)) output_dims = [] for (idx, input_dim) in enumerate(input_dims): output_dim = ((((input_dim + (2 * paddings[idx])) - ((dilations[idx] * (kernel_dims[idx] - 1)) + 1)) // strides[idx]) + 1) output_dims.append(output_dim) filters_per_channel = (out_channels // groups) conv_per_position_macs = ((int(_prod(kernel_dims)) * in_channels) * filters_per_channel) active_elements_count = (batch_size * int(_prod(output_dims))) overall_conv_macs = (conv_per_position_macs * active_elements_count) overall_conv_flops = (2 * overall_conv_macs) bias_flops = 0 if (bias is not None): bias_flops = (out_channels * active_elements_count) return (int((overall_conv_flops + bias_flops)), int(overall_conv_macs))
class AdversarialEpocher(SemiSupervisedEpocher, ABC): def _assertion(self): pass def __init__(self, *, model: nn.Module, optimizer: T_optim, labeled_loader: T_loader, unlabeled_loader: T_loader, sup_criterion: T_loss, num_batches: int, cur_epoch=0, device='cpu', two_stage: bool=False, disable_bn: bool=False, discriminator=None, discr_optimizer=None, reg_weight=None, dis_consider_image: bool, **kwargs) -> None: super().__init__(model=model, optimizer=optimizer, labeled_loader=labeled_loader, unlabeled_loader=unlabeled_loader, sup_criterion=sup_criterion, num_batches=num_batches, cur_epoch=cur_epoch, device=device, two_stage=two_stage, disable_bn=disable_bn, **kwargs) self._discriminator = discriminator self._discr_optimizer = discr_optimizer self._reg_weight = float(reg_weight) assert isinstance(discriminator, nn.Module) assert isinstance(discr_optimizer, torch.optim.Optimizer) self._dis_consider_image = dis_consider_image def _run(self, **kwargs): self.meters['lr'].add(get_lrs_from_optimizer(self._optimizer)) self._model.train() return self._run_adver(**kwargs) def configure_meters(self, meters: MeterInterface) -> MeterInterface: meters = super(AdversarialEpocher, self).configure_meters(meters) meters.delete_meter('reg_loss') with self.meters.focus_on('adv_reg'): meters.register_meter('dis_loss', AverageValueMeter()) meters.register_meter('gen_loss', AverageValueMeter()) meters.register_meter('reg_weight', AverageValueMeter()) return meters def _run_adver(self, **kwargs): criterion = nn.BCELoss() TRUE_LABEL = 1.0 FAKE_LABEL = 0.0 optimizerD = self._discr_optimizer optimizerG = self._optimizer with self.meters.focus_on('adv_reg'): self.meters['reg_weight'].add(self._reg_weight) for (self.cur_batch_num, labeled_data) in zip(self.indicator, self._labeled_loader): ((labeled_image, _), labeled_target, labeled_filename, _, label_group) = self._unzip_data(labeled_data, self._device) if (self._reg_weight > 0): unlabeled_data = next(self.unlabeled_iter) ((unlabeled_image, _), _, unlabeled_filename, unl_partition, unl_group) = self._unzip_data(unlabeled_data, self._device) if (self.cur_batch_num < 5): if (self._reg_weight > 0): logger.trace(f"{self.__class__.__name__}--cur_batch:{self.cur_batch_num}, labeled_filenames: {','.join(labeled_filename)}, unlabeled_filenames: {','.join(unlabeled_filename)}") else: logger.trace(f"{self.__class__.__name__}--cur_batch:{self.cur_batch_num}, labeled_filenames: {','.join(labeled_filename)}") self._optimizer.zero_grad() labeled_logits = self._model(labeled_image) onehot_target = class2one_hot(labeled_target.squeeze(1), self.num_classes) sup_loss = self._sup_criterion(labeled_logits.softmax(1), onehot_target) generator_err = torch.tensor(0, device=self.device, dtype=torch.float) if (self._reg_weight > 0): unlabeled_logits = self._model(unlabeled_image) if self._dis_consider_image: discr_output_unlabeled = self._discriminator(torch.cat([unlabeled_image, unlabeled_logits.softmax(1)], dim=1)) else: discr_output_unlabeled = self._discriminator(unlabeled_logits.softmax(1)) generator_err = criterion(discr_output_unlabeled, torch.zeros_like(discr_output_unlabeled).fill_(TRUE_LABEL)) generator_loss = (sup_loss + (self._reg_weight * generator_err)) generator_loss.backward() optimizerG.step() if self.on_master(): with torch.no_grad(): self.meters['sup_loss'].add(sup_loss.item()) self.meters['sup_dice'].add(labeled_logits.max(1)[1], labeled_target.squeeze(1), group_name=label_group) with self.meters.focus_on('adv_reg'): self.meters['gen_loss'].add(generator_err.item()) disc_loss = torch.tensor(0, device=self.device, dtype=torch.float) if (self._reg_weight > 0): self._discriminator.zero_grad() if self._dis_consider_image: discr_output_labeled = self._discriminator(torch.cat([labeled_image, labeled_logits.detach().softmax(1)], dim=1)) else: discr_output_labeled = self._discriminator(labeled_logits.detach().softmax(1)) discr_err_labeled = criterion(discr_output_labeled, torch.zeros_like(discr_output_labeled).fill_(TRUE_LABEL)) if self._dis_consider_image: discr_output_unlabeled = self._discriminator(torch.cat([unlabeled_image, unlabeled_logits.detach().softmax(1)], dim=1)) else: discr_output_unlabeled = self._discriminator(unlabeled_logits.detach().softmax(1)) discr_err_unlabeled = criterion(discr_output_unlabeled, torch.zeros_like(discr_output_unlabeled).fill_(FAKE_LABEL)) disc_loss = (discr_err_labeled + discr_err_unlabeled) (disc_loss * self._reg_weight).backward() optimizerD.step() if self.on_master(): with self.meters.focus_on('adv_reg'): self.meters['dis_loss'].add(disc_loss.item()) report_dict = self.meters.statistics() self.indicator.set_postfix_statics(report_dict, cache_time=10) _cache() def unlabeled_iter(self): return iter(self._unlabeled_loader)
class TestCommutationAnalysis(QiskitTestCase): def setUp(self): self.pass_ = CommutationAnalysis() self.pset = self.pass_.property_set = PropertySet() def assertCommutationSet(self, result, expected): result_to_compare = {} for (qbit_str, sets) in result.items(): if (not isinstance(qbit_str, str)): continue result_to_compare[qbit_str] = [] for commutation_set in sets: result_to_compare[qbit_str].append(sorted([node._node_id for node in commutation_set])) for (qbit_str, sets) in expected.items(): for commutation_set in sets: commutation_set.sort() self.assertDictEqual(result_to_compare, expected) def test_commutation_set_property_is_created(self): qr = QuantumRegister(3, 'qr') circuit = QuantumCircuit(qr) circuit.h(qr) dag = circuit_to_dag(circuit) self.assertIsNone(self.pset['commutation_set']) self.pass_.run(dag) self.assertIsNotNone(self.pset['commutation_set']) def test_all_gates(self): qr = QuantumRegister(2, 'qr') circuit = QuantumCircuit(qr) circuit.h(qr[0]) circuit.x(qr[0]) circuit.y(qr[0]) circuit.t(qr[0]) circuit.s(qr[0]) circuit.rz(0.5, qr[0]) circuit.u1(0.5, qr[0]) circuit.u2(0.5, 0.6, qr[0]) circuit.u3(0.5, 0.6, 0.7, qr[0]) circuit.cx(qr[0], qr[1]) circuit.cy(qr[0], qr[1]) circuit.cz(qr[0], qr[1]) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [5], [6], [7], [8, 9, 10, 11], [12], [13], [14], [15], [16], [2]], 'qr[1]': [[3], [14], [15], [16], [4]]} self.assertCommutationSet(self.pset['commutation_set'], expected) def test_non_commutative_circuit(self): qr = QuantumRegister(3, 'qr') circuit = QuantumCircuit(qr) circuit.h(qr) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [7], [2]], 'qr[1]': [[3], [8], [4]], 'qr[2]': [[5], [9], [6]]} self.assertCommutationSet(self.pset['commutation_set'], expected) def test_non_commutative_circuit_2(self): qr = QuantumRegister(3, 'qr') circuit = QuantumCircuit(qr) circuit.cx(qr[0], qr[1]) circuit.h(qr[2]) circuit.cx(qr[1], qr[2]) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [7], [2]], 'qr[1]': [[3], [7], [9], [4]], 'qr[2]': [[5], [8], [9], [6]]} self.assertCommutationSet(self.pset['commutation_set'], expected) def test_commutative_circuit(self): qr = QuantumRegister(3, 'qr') circuit = QuantumCircuit(qr) circuit.cx(qr[0], qr[1]) circuit.h(qr[2]) circuit.cx(qr[2], qr[1]) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [7], [2]], 'qr[1]': [[3], [7, 9], [4]], 'qr[2]': [[5], [8], [9], [6]]} self.assertCommutationSet(self.pset['commutation_set'], expected) def test_commutative_circuit_2(self): qr = QuantumRegister(3, 'qr') circuit = QuantumCircuit(qr) circuit.cx(qr[0], qr[1]) circuit.z(qr[0]) circuit.h(qr[2]) circuit.cx(qr[2], qr[1]) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [7, 8], [2]], 'qr[1]': [[3], [7, 10], [4]], 'qr[2]': [[5], [9], [10], [6]]} self.assertCommutationSet(self.pset['commutation_set'], expected) def test_commutative_circuit_3(self): qr = QuantumRegister(3, 'qr') circuit = QuantumCircuit(qr) circuit.cx(qr[0], qr[1]) circuit.h(qr[2]) circuit.z(qr[0]) circuit.cx(qr[2], qr[1]) circuit.cx(qr[0], qr[1]) circuit.x(qr[2]) circuit.z(qr[0]) circuit.cx(qr[1], qr[2]) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [7, 9, 11, 13], [2]], 'qr[1]': [[3], [7, 10, 11], [14], [4]], 'qr[2]': [[5], [8], [10], [12, 14], [6]]} self.assertCommutationSet(self.pset['commutation_set'], expected) def test_jordan_wigner_type_circuit(self): qr = QuantumRegister(6, 'qr') circuit = QuantumCircuit(qr) circuit.cx(qr[0], qr[1]) circuit.cx(qr[1], qr[2]) circuit.cx(qr[2], qr[3]) circuit.cx(qr[3], qr[4]) circuit.cx(qr[4], qr[5]) circuit.z(qr[5]) circuit.cx(qr[4], qr[5]) circuit.cx(qr[3], qr[4]) circuit.cx(qr[2], qr[3]) circuit.cx(qr[1], qr[2]) circuit.cx(qr[0], qr[1]) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [13, 23], [2]], 'qr[1]': [[3], [13], [14, 22], [23], [4]], 'qr[2]': [[5], [14], [15, 21], [22], [6]], 'qr[3]': [[7], [15], [16, 20], [21], [8]], 'qr[4]': [[9], [16], [17, 19], [20], [10]], 'qr[5]': [[11], [17], [18], [19], [12]]} self.assertCommutationSet(self.pset['commutation_set'], expected) def test_all_commute_circuit(self): qr = QuantumRegister(5, 'qr') circuit = QuantumCircuit(qr) circuit.cx(qr[0], qr[1]) circuit.cx(qr[2], qr[1]) circuit.cx(qr[4], qr[3]) circuit.cx(qr[2], qr[3]) circuit.z(qr[0]) circuit.z(qr[4]) circuit.cx(qr[0], qr[1]) circuit.cx(qr[2], qr[1]) circuit.cx(qr[4], qr[3]) circuit.cx(qr[2], qr[3]) dag = circuit_to_dag(circuit) self.pass_.run(dag) expected = {'qr[0]': [[1], [11, 15, 17], [2]], 'qr[1]': [[3], [11, 12, 17, 18], [4]], 'qr[2]': [[5], [12, 14, 18, 20], [6]], 'qr[3]': [[7], [13, 14, 19, 20], [8]], 'qr[4]': [[9], [13, 16, 19], [10]]} self.assertCommutationSet(self.pset['commutation_set'], expected)
class RandomColorDistortion(): def __init__(self, s: float=1.0): self.color_distort = compose_color_distortion(s=s) def __call__(self, x): return self.color_distort(x)
def train(cfg, local_rank, distributed, tblogger=None, transfer_weight=False, adjust_lr=False, skip_val=False, no_head=False): model = build_detection_model(cfg) device = torch.device('cuda') model.to(device) optimizer = make_optimizer(cfg, model) scheduler = make_lr_scheduler(cfg, optimizer) if distributed: model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[local_rank], output_device=local_rank, broadcast_buffers=False, find_unused_parameters=True) arguments = {} arguments['iteration'] = 0 arguments['person_pool'] = MemoryPool() output_dir = cfg.OUTPUT_DIR save_to_disk = (get_rank() == 0) checkpointer = ActionCheckpointer(cfg, model, optimizer, scheduler, output_dir, save_to_disk) extra_checkpoint_data = checkpointer.load(cfg.MODEL.WEIGHT, model_weight_only=transfer_weight, adjust_scheduler=adjust_lr, no_head=no_head) arguments.update(extra_checkpoint_data) data_loader = make_data_loader(cfg, is_train=True, is_distributed=distributed, start_iter=arguments['iteration']) checkpoint_period = cfg.SOLVER.CHECKPOINT_PERIOD val_period = cfg.SOLVER.EVAL_PERIOD mem_active = has_memory(cfg.MODEL.HIT_STRUCTURE) if (not skip_val): dataset_names_val = cfg.DATASETS.TEST data_loaders_val = make_data_loader(cfg, is_train=False, is_distributed=distributed) else: dataset_names_val = [] data_loaders_val = [] do_train(model, data_loader, optimizer, scheduler, checkpointer, device, checkpoint_period, arguments, tblogger, val_period, dataset_names_val, data_loaders_val, distributed, mem_active) return model
class Discriminator(nn.Module): def __init__(self, preprocess_GAN_mode, input_channel, batch_size=64, image_size=64, conv_dim=64): super(Discriminator, self).__init__() self.imsize = image_size layer1 = [] layer2 = [] layer3 = [] last = [] layer1.append(SpectralNorm(nn.Conv2d(input_channel, conv_dim, 4, 2, 1))) layer1.append(nn.LeakyReLU(0.1)) curr_dim = conv_dim layer2.append(SpectralNorm(nn.Conv2d(curr_dim, (curr_dim * 2), 4, 2, 1))) layer2.append(nn.LeakyReLU(0.1)) curr_dim = (curr_dim * 2) layer3.append(SpectralNorm(nn.Conv2d(curr_dim, (curr_dim * 2), 4, 2, 1))) layer3.append(nn.LeakyReLU(0.1)) curr_dim = (curr_dim * 2) if (self.imsize == 65): layer4 = [] layer4.append(SpectralNorm(nn.Conv2d(curr_dim, (curr_dim * 2), 4, 2, 1))) layer4.append(nn.LeakyReLU(0.1)) self.l4 = nn.Sequential(*layer4) curr_dim = (curr_dim * 2) self.l1 = nn.Sequential(*layer1) self.l2 = nn.Sequential(*layer2) self.l3 = nn.Sequential(*layer3) last.append(nn.Conv2d(curr_dim, 1, 4)) self.last = nn.Sequential(*last) self.attn1 = Self_Attn(256, 'relu') self.attn2 = Self_Attn(512, 'relu') if (preprocess_GAN_mode == 1): self.preprocess_additional = nn.BatchNorm2d(input_channel) elif (preprocess_GAN_mode == 2): self.preprocess_additional = nn.Tanh() elif (preprocess_GAN_mode == 3): self.preprocess_additional = (lambda x: (2 * ((x / 255) - 0.5))) else: raise ValueError('preprocess_GAN_mode should be 1:bn or 2:tanh or 3:-1 - 1') def forward(self, x): x = self.preprocess_additional(x) out = self.l1(x) out = self.l2(out) out = self.l3(out) (out, p1) = self.attn1(out) out = self.l4(out) (out, p2) = self.attn2(out) out = self.last(out) return [out, p1, p2]
def custom_figure(n_panels=2, width=8.0, panel_aspect_ratio=1.0, extra_top_space=False, reduce_vertical_sep=False): if isinstance(n_panels, collections.Sequence): (n_panels_h, n_panels_v) = n_panels else: n_panels_h = n_panels n_panels_v = 1 _margin_t_absolute = (margin_t_absolute_extra if extra_top_space else margin_t_absolute) margin_l = (margin_l_absolute / width) margin_r = (margin_r_absolute / width) margin_l_subsequent = margin_l if (n_panels_h > 2): margin_l_subsequent = margin_r margin_sep = (margin_sep_absolute / width) if (n_panels_h > 2): margin_sep = 0 margin_sep_total = ((margin_r + margin_sep) + margin_l_subsequent) panel_width = ((((1.0 - margin_l) - margin_r) - ((n_panels_h - 1) * margin_sep_total)) / n_panels_h) wspace = (margin_sep_total / panel_width) panel_height_absolute = ((panel_width * width) / panel_aspect_ratio) height = ((n_panels_v * ((panel_height_absolute + _margin_t_absolute) + margin_b_absolute)) + ((n_panels_v - 1) * margin_sep_absolute)) panel_height = (panel_height_absolute / height) margin_t = (_margin_t_absolute / height) margin_b = (margin_b_absolute / height) if reduce_vertical_sep: margin_sep_total = (margin_sep_small_absolute / height) else: margin_sep_total = ((margin_t + margin_b) + (margin_sep_absolute / height)) hspace = (margin_sep_total / panel_height) fig = plt.figure(figsize=(width, height)) plt.subplots_adjust(left=margin_l, right=(1.0 - margin_r), bottom=margin_b, top=(1.0 - margin_t), wspace=wspace, hspace=hspace) return fig
def get_fbank(path_or_fp: Union[(str, BinaryIO)], n_bins=80) -> np.ndarray: (sound, sample_rate) = get_waveform(path_or_fp, normalization=False) features = _get_kaldi_fbank(sound, sample_rate, n_bins) if (features is None): features = _get_torchaudio_fbank(sound, sample_rate, n_bins) if (features is None): raise ImportError('Please install pyKaldi or torchaudio to enable online filterbank feature extraction') return features
def load_classifier(name='resnet101', n=2): model = torchvision.models.__dict__[name](pretrained=True) filters = model.fc.weight.shape[1] model.fc.bias = nn.Parameter(torch.zeros(n), requires_grad=True) model.fc.weight = nn.Parameter(torch.zeros(n, filters), requires_grad=True) model.fc.out_features = n return model
def initialize_hyperparameters(PATHS: dict, load_target: str, config_name: str='default', n_envs: int=1) -> dict: if (load_target is None): hyperparams = load_hyperparameters_json(PATHS=PATHS, from_scratch=True, config_name=config_name) hyperparams['agent_name'] = PATHS['model'].split('/')[(- 1)] else: hyperparams = load_hyperparameters_json(PATHS=PATHS) check_batch_size(n_envs, hyperparams['batch_size'], hyperparams['m_batch_size']) hyperparams['n_steps'] = int((hyperparams['batch_size'] / n_envs)) write_hyperparameters_json(hyperparams, PATHS) print_hyperparameters(hyperparams) return hyperparams
def timed_run(f: FunctionType, timeout_seconds: int=3600) -> Tuple[(Any, Number, bool)]: start_time = time.time() with Timeout(timeout_seconds) as timeout_ctx: res = f() duration = (time.time() - start_time) timed_out = (timeout_ctx.state == timeout_ctx.TIMED_OUT) if timed_out: res = None return (res, duration, timed_out)
_arguments_as_properties('lost_lang', 'known_lang', 'capacity', 'num_cognates') class Evaluator(): def __init__(self, model, data_loader): self.model = model self.data_loader = data_loader self._settings = list() def add_setting(self, mode=None, edit=None): assert (mode in ['mle', 'flow']) assert (edit in [True, False]) lost_size = self.data_loader.size(self.lost_lang) known_size = self.data_loader.size(self.known_lang) if (mode == 'mle'): self._settings.append(EvalSetting(self.lost_lang, self.known_lang, lost_size, known_size, mode, None, None)) else: for c in self.capacity: self._settings.append(EvalSetting(self.lost_lang, self.known_lang, lost_size, known_size, mode, edit, c)) def __str__(self): table = pt() table.field_names = ('lost', 'known', 'lost_size', 'known_size', 'mode', 'edit', 'capacity') for s in self._settings: table.add_row([getattr(s, field) for field in table.field_names]) table.align = 'l' return str(table) def evaluate(self, epoch, num_cognates): self.model.eval() table = pt() table.field_names = ('lost', 'known', 'mode', 'edit', 'capacity', 'score') eval_scores = dict() for s in self._settings: batch = self.data_loader.entire_batch model_ret = self.model(batch, mode=s.mode, num_cognates=num_cognates, edit=s.edit, capacity=s.capacity) almt = (model_ret.valid_log_probs if (s.mode == 'mle') else model_ret.flow) preds = almt.get_best() acc = self._evaluate_one_setting(preds) score = (acc / len(preds)) fmt_score = f'{acc}/{len(preds)}={score:.3f}' table.add_row(([getattr(s, field) for field in table.field_names[:(- 1)]] + [fmt_score])) eval_scores[str(s)] = score table.align = 'l' table.title = f'Epoch: {epoch}' log_pp(table) return eval_scores def _evaluate_one_setting(self, preds): acc = 0 for (lost, known) in preds.items(): if is_cognate(lost, known): acc += 1 return acc
def _handle_path(path, sess, low_profile=False): if isinstance(path, str): f = np.load(path) (m, s) = (f['mu'][:], f['sigma'][:]) f.close() else: files = path if low_profile: (m, s) = calculate_activation_statistics_from_files(files, sess) else: x = path (m, s) = calculate_activation_statistics(x, sess) del x return (m, s)
def num_dependent_clauses(const_pt): dep_clauses = [] clause_tags = None if (settings.LANGUAGE in ['zh-hant', 'fr']): lang = settings.LANGUAGE else: lang = 'default' clause_tags = SUBORD_CLAUSE_LANGUAGE_MAP[lang] for clause_tag in clause_tags: for leaf in _leaves(const_pt, clause_tag): dep_clauses.append(leaf.leaves()) return len(dep_clauses)
class ConfigTester(unittest.TestCase): def test_outputs_single_attribute(self): outputs = CustomOutput(images=np.random.rand(1, 3, 4, 4)) assert isinstance(outputs.images, np.ndarray) assert (outputs.images.shape == (1, 3, 4, 4)) assert isinstance(outputs['images'], np.ndarray) assert (outputs['images'].shape == (1, 3, 4, 4)) assert isinstance(outputs[0], np.ndarray) assert (outputs[0].shape == (1, 3, 4, 4)) outputs = CustomOutput(images=[PIL.Image.new('RGB', (4, 4))]) assert isinstance(outputs.images, list) assert isinstance(outputs.images[0], PIL.Image.Image) assert isinstance(outputs['images'], list) assert isinstance(outputs['images'][0], PIL.Image.Image) assert isinstance(outputs[0], list) assert isinstance(outputs[0][0], PIL.Image.Image) def test_outputs_dict_init(self): outputs = CustomOutput({'images': np.random.rand(1, 3, 4, 4)}) assert isinstance(outputs.images, np.ndarray) assert (outputs.images.shape == (1, 3, 4, 4)) assert isinstance(outputs['images'], np.ndarray) assert (outputs['images'].shape == (1, 3, 4, 4)) assert isinstance(outputs[0], np.ndarray) assert (outputs[0].shape == (1, 3, 4, 4)) outputs = CustomOutput({'images': [PIL.Image.new('RGB', (4, 4))]}) assert isinstance(outputs.images, list) assert isinstance(outputs.images[0], PIL.Image.Image) assert isinstance(outputs['images'], list) assert isinstance(outputs['images'][0], PIL.Image.Image) assert isinstance(outputs[0], list) assert isinstance(outputs[0][0], PIL.Image.Image) def test_outputs_serialization(self): outputs_orig = CustomOutput(images=[PIL.Image.new('RGB', (4, 4))]) serialized = pkl.dumps(outputs_orig) outputs_copy = pkl.loads(serialized) assert (dir(outputs_orig) == dir(outputs_copy)) assert (dict(outputs_orig) == dict(outputs_copy)) assert (vars(outputs_orig) == vars(outputs_copy)) _torch def test_torch_pytree(self): import torch import torch.utils._pytree data = np.random.rand(1, 3, 4, 4) x = CustomOutput(images=data) self.assertFalse(torch.utils._pytree._is_leaf(x)) expected_flat_outs = [data] expected_tree_spec = torch.utils._pytree.TreeSpec(CustomOutput, ['images'], [torch.utils._pytree.LeafSpec()]) (actual_flat_outs, actual_tree_spec) = torch.utils._pytree.tree_flatten(x) self.assertEqual(expected_flat_outs, actual_flat_outs) self.assertEqual(expected_tree_spec, actual_tree_spec) unflattened_x = torch.utils._pytree.tree_unflatten(actual_flat_outs, actual_tree_spec) self.assertEqual(x, unflattened_x)
def update_config(config, data_sets): config.max_num_sents = 0 config.max_sent_size = 0 config.max_ques_size = 0 config.max_ques_sub_size = 0 config.max_word_size = 0 config.max_para_size = 0 for data_set in data_sets: data = data_set.data shared = data_set.shared for idx in data_set.valid_idxs: q = data['q'][idx] sents = data['x'][idx] config.max_para_size = max(config.max_para_size, sum(map(len, sents))) config.max_num_sents = max(config.max_num_sents, len(sents)) config.max_sent_size = max(config.max_sent_size, max(map(len, sents))) config.max_word_size = max(config.max_word_size, max((len(word) for sent in sents for word in sent))) if (len(q) > 0): config.max_ques_size = max(config.max_ques_size, len(q)) config.max_word_size = max(config.max_word_size, max((len(word) for word in q))) if (config.mode == 'train'): config.max_num_sents = min(config.max_num_sents, config.num_sents_th) config.max_sent_size = min(config.max_sent_size, config.sent_size_th) config.max_para_size = min(config.max_para_size, config.para_size_th) config.max_word_size = min(config.max_word_size, config.word_size_th) config.char_vocab_size = len(data_sets[0].shared['char2idx']) config.word_emb_size = len(next(iter(data_sets[0].shared['word2vec'].values()))) config.word_vocab_size = len(data_sets[0].shared['word2idx']) if config.single: config.max_num_sents = 1 if config.squash: config.max_sent_size = config.max_para_size config.max_num_sents = 1
def render_batch(visualize_fn, input, target, output): batch_size = input.shape[0] (fig, axes) = plt.subplots(nrows=batch_size, ncols=3, figsize=(12, (4 * batch_size))) plt.subplots_adjust(left=0.05, bottom=0, right=0.95, top=1, hspace=0) for i in range(batch_size): ax = (axes if (batch_size == 1) else axes[i]) ax[0].imshow(input[i]) visualize_fn(ax[1], input[i], target[i]) visualize_fn(ax[2], input[i], output[i]) fig.canvas.draw() data = (np.fromstring(fig.canvas.tostring_rgb(), dtype=np.uint8, sep='') / 255.0) data = data.reshape((fig.canvas.get_width_height()[::(- 1)] + (3,))) plt.close(fig) return data
def create_model(arch, heads, head_conv): num_layers = (int(arch[(arch.find('_') + 1):]) if ('_' in arch) else 0) arch = (arch[:arch.find('_')] if ('_' in arch) else arch) get_model = _model_factory[arch] model = get_model(num_layers=num_layers, heads=heads, head_conv=head_conv) return model
def labels2clusters(labels): lb2idxs = {} for (idx, lb) in enumerate(labels): if (lb not in lb2idxs): lb2idxs[lb] = [] lb2idxs[lb].append(idx) clusters = [idxs for (_, idxs) in lb2idxs.items()] return clusters
_tf class TFCTRLModelTest(TFModelTesterMixin, PipelineTesterMixin, unittest.TestCase): all_model_classes = ((TFCTRLModel, TFCTRLLMHeadModel, TFCTRLForSequenceClassification) if is_tf_available() else ()) all_generative_model_classes = ((TFCTRLLMHeadModel,) if is_tf_available() else ()) pipeline_model_mapping = ({'feature-extraction': TFCTRLModel, 'text-classification': TFCTRLForSequenceClassification, 'text-generation': TFCTRLLMHeadModel, 'zero-shot': TFCTRLForSequenceClassification} if is_tf_available() else {}) test_head_masking = False test_onnx = False def is_pipeline_test_to_skip(self, pipeline_test_casse_name, config_class, model_architecture, tokenizer_name, processor_name): if (pipeline_test_casse_name == 'ZeroShotClassificationPipelineTests'): return True return False def setUp(self): self.model_tester = TFCTRLModelTester(self) self.config_tester = ConfigTester(self, config_class=CTRLConfig, n_embd=37) def test_config(self): self.config_tester.run_common_tests() def test_ctrl_model(self): config_and_inputs = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_ctrl_model(*config_and_inputs) def test_ctrl_lm_head(self): config_and_inputs = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_ctrl_lm_head(*config_and_inputs) def test_ctrl_sequence_classification_model(self): config_and_inputs = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_ctrl_for_sequence_classification(*config_and_inputs) def test_model_common_attributes(self): (config, inputs_dict) = self.model_tester.prepare_config_and_inputs_for_common() list_lm_models = [TFCTRLLMHeadModel] list_other_models_with_output_ebd = [TFCTRLForSequenceClassification] for model_class in self.all_model_classes: model = model_class(config) assert isinstance(model.get_input_embeddings(), tf.keras.layers.Layer) if (model_class in list_lm_models): x = model.get_output_embeddings() assert isinstance(x, tf.keras.layers.Layer) name = model.get_bias() assert isinstance(name, dict) for (k, v) in name.items(): assert isinstance(v, tf.Variable) elif (model_class in list_other_models_with_output_ebd): x = model.get_output_embeddings() assert isinstance(x, tf.keras.layers.Layer) name = model.get_bias() assert (name is None) else: x = model.get_output_embeddings() assert (x is None) name = model.get_bias() assert (name is None) def test_model_from_pretrained(self): for model_name in TF_CTRL_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: model = TFCTRLModel.from_pretrained(model_name) self.assertIsNotNone(model)
def summarize_evaluation(evaluation: pd.DataFrame) -> dict: if ((evaluation is None) or (len(evaluation) == 0)): warnings.warn('No completions to evaluate.') return None return {'accuracy': evaluation.correct.mean(), 'contains_answer': evaluation.contains_answer.mean(), 'correct_format': evaluation.correct_format.mean(), 'complete': evaluation.complete.mean()}
class HistoricalContainer(metaclass=ABCMeta): def __init__(self) -> None: self._record_dict: _Save_Type = OrderedDict() self._current_epoch: int = 0 def __enter__(self): return self def __exit__(self, exc_type, exc_val, exc_tb): pass def trainer(self): return self._trainer def record_dict(self) -> _Save_Type: return self._record_dict def get_record_dict(self, epoch=None) -> Union[(_Record_Type, _Save_Type)]: if (epoch is None): return self.record_dict assert (epoch in self._record_dict.keys()), 'epoch {} not saved in {}'.format(epoch, ', '.join(list(self._record_dict.keys()))) return self.record_dict[epoch] def current_epoch(self) -> int: return self._current_epoch def summary(self) -> pd.DataFrame: validated_table = pd.DataFrame(self.record_dict).T if (len(self.record_dict) < self.current_epoch): missing_table = pd.DataFrame(index=(set(range(self.current_epoch)) - set(self.record_dict.keys()))) validated_table = validated_table.append(missing_table, sort=True) return validated_table def add(self, input_dict: _Record_Type, epoch=None) -> None: if epoch: self._current_epoch = epoch self._record_dict[self._current_epoch] = input_dict self._current_epoch += 1 def reset(self) -> None: self._record_dict: _Save_Type = OrderedDict() self._current_epoch = 0 def state_dict(self) -> Dict[(str, Any)]: return self.__dict__ def load_state_dict(self, state_dict: Dict[(str, Any)]) -> None: self.__dict__.update(state_dict) def __repr__(self): return str(pd.DataFrame(self.record_dict).T)
def write_podspec(f, rules, args): rule_dir = build_rule_directory(rules)['abseil'] spec = re.sub('\\$\\{(\\w+)\\}', (lambda x: args[x.group(1)]), SPEC_TEMPLATE).lstrip() f.write(spec) write_podspec_map(f, rule_dir, 0) f.write('end\n')
def add_ground_truth_to_proposals_single_image(gt_boxes, proposals): device = proposals.objectness_logits.device gt_logit_value = math.log(((1.0 - 1e-10) / (1 - (1.0 - 1e-10)))) gt_logits = (gt_logit_value * torch.ones(len(gt_boxes), device=device)) gt_proposal = Instances(proposals.image_size) gt_proposal.proposal_boxes = gt_boxes gt_proposal.objectness_logits = gt_logits new_proposals = Instances.cat([proposals, gt_proposal]) return new_proposals
class Up(nn.Module): def __init__(self, in_channels, out_channels, bilinear=True): super().__init__() if bilinear: self.up = nn.Upsample(scale_factor=2, mode='bilinear', align_corners=True) self.conv = DoubleConv(in_channels, out_channels, (in_channels // 2)) else: self.up = nn.ConvTranspose2d(in_channels, (in_channels // 2), kernel_size=2, stride=2) self.conv = DoubleConv(in_channels, out_channels) def forward(self, x1, x2): x1 = self.up(x1) diffY = (x2.size()[2] - x1.size()[2]) diffX = (x2.size()[3] - x1.size()[3]) x1 = F.pad(x1, [(diffX // 2), (diffX - (diffX // 2)), (diffY // 2), (diffY - (diffY // 2))]) x = torch.cat([x2, x1], dim=1) return self.conv(x)
class TestTorchOP(unittest.TestCase): def setUpClass(self): pass def tearDownClass(self): pass def test_1(self): n = Net('div') example_in = torch.rand(3, 256) example_in2 = torch.rand(256) traced_model = torch.jit.trace(n, (example_in, example_in2)) torch.jit.save(traced_model, '{}.pt'.format(file_name)) ref_out = traced_model(example_in, example_in2).detach().numpy() graph = compile('{}.pt'.format(file_name)) graph.save(file_name) newgraph = Graph() newgraph.graph_init((file_name + '/conf.yaml'), (file_name + '/model.bin')) out = newgraph.inference([example_in.numpy(), example_in2.numpy()]) np.testing.assert_almost_equal(ref_out, [*out.values()][0], decimal=5) os.remove('{}.pt'.format(file_name)) shutil.rmtree(file_name) def test_2(self): n = Net('mul') example_in = torch.rand(3, 256) example_in2 = torch.rand(256) traced_model = torch.jit.trace(n, (example_in, example_in2)) torch.jit.save(traced_model, '{}.pt'.format(file_name)) ref_out = traced_model(example_in, example_in2).detach().numpy() graph = compile('{}.pt'.format(file_name)) graph.save(file_name) newgraph = Graph() newgraph.graph_init((file_name + '/conf.yaml'), (file_name + '/model.bin')) out = newgraph.inference([example_in.numpy(), example_in2.numpy()]) np.testing.assert_almost_equal(ref_out, [*out.values()][0], decimal=5) os.remove('{}.pt'.format(file_name)) shutil.rmtree(file_name)
def score_amr_pairs(f1, f2, justinstance=False, justattribute=False, justrelation=False): total_match_num = total_test_num = total_gold_num = 0 for (sent_num, (cur_amr1, cur_amr2)) in enumerate(generate_amr_lines(f1, f2), start=1): (best_match_num, test_triple_num, gold_triple_num) = get_amr_match(cur_amr1, cur_amr2, sent_num=sent_num, justinstance=justinstance, justattribute=justattribute, justrelation=justrelation) total_match_num += best_match_num total_test_num += test_triple_num total_gold_num += gold_triple_num match_triple_dict.clear() if (not single_score): (yield compute_f(best_match_num, test_triple_num, gold_triple_num)) if verbose: print('Total match number, total triple number in AMR 1, and total triple number in AMR 2:', file=DEBUG_LOG) print(total_match_num, total_test_num, total_gold_num, file=DEBUG_LOG) print('', file=DEBUG_LOG) if single_score: (yield compute_f(total_match_num, total_test_num, total_gold_num))
class YGate(Gate): def __init__(self, label=None): super().__init__('y', 1, [], label=label) def _define(self): definition = [] q = QuantumRegister(1, 'q') rule = [(U3Gate(pi, (pi / 2), (pi / 2)), [q[0]], [])] for inst in rule: definition.append(inst) self.definition = definition def inverse(self): return YGate() def to_matrix(self): return numpy.array([[0, (- 1j)], [1j, 0]], dtype=complex)
def get_target_feature(model, preprocess, tokenizer_funct, device, target_images=None, target_prompts=None): if (target_images is not None): with torch.no_grad(): curr_images = [preprocess(i).unsqueeze(0) for i in target_images] curr_images = torch.concatenate(curr_images).to(device) all_target_features = model.encode_image(curr_images) else: texts = tokenizer_funct(target_prompts).to(device) all_target_features = model.encode_text(texts) return all_target_features
def main(): with codecs.open('results.csv', 'w', 'utf-8') as fp: writer = csv.writer(fp) writer.writerow(['experiment', 'model', 'error', 'elapsed']) start_time = time.time() run_experiment(writer, 'counts', generate_data_counts) run_experiment(writer, 'quad', generate_data_quad) run_experiment(writer, 'sqrt', generate_data_fn(1, 1.0, 100.0, math.sqrt)) run_experiment(writer, 'log', generate_data_fn(1, 1.0, 100.0, math.log)) run_experiment(writer, 'pow', generate_data_fn(1, 1.0, 10.0, (lambda x: (x ** 2)))) run_experiment(writer, 'ratio', generate_data_ratio) run_experiment(writer, 'diff', generate_data_diff) run_experiment(writer, 'r_poly', generate_data_fn(1, 1.0, 10.0, (lambda x: (1 / ((5 * x) + (8 * (x ** 2))))))) run_experiment(writer, 'poly', generate_data_fn(1, 0.0, 2.0, (lambda x: ((1 + (5 * x)) + (8 * (x ** 2)))))) run_experiment(writer, 'r_diff', generate_data_fn(4, 1.0, 10.0, (lambda a, b, c, d: ((a - b) / (c - d))))) elapsed_time = (time.time() - start_time) print('Elapsed time: {}'.format(hms_string(elapsed_time)))
class FangraphsBattingStats(FangraphsStatsBase): COMMON = 'c' LINE_BREAK = '-1' NAME = '0' TEAM = '1' SEASON = '2' AGE = '3' G = '4' GAMES = G AB = '5' AT_BATS = AB PA = '6' PLATE_APPEARANCES = PA H = '7' HITS = H SINGLES = '8' DOUBLES = '9' TRIPLES = '10' HR = '11' HOME_RUNS = HR R = '12' RUNS = R RBI = '13' BB = '14' WALKS = BB IBB = '15' INTENTIONAL_WALKS = IBB SO = '16' STRIKE_OUTS = SO HBP = '17' HIT_BY_PITCH = HBP SF = '18' SACRIFICE_FLY = SF SH = '19' SACRIFICE_HIT = SH GDP = '20' GROUNDED_DOUBLE_PLAY = GDP SB = '21' STOLEN_BASES = SB CS = '22' CAUGHT_STEALING = CS AVG = '23' BATTING_AVERAGE = AVG GB = '24' GROUND_BALLS = GB FB = '25' FLY_BALLS = FB LD = '26' IFFB = '27' PITCHES = '28' BALLS = '29' STRIKES = '30' IFH = '31' BU = '32' BUH = '33' BB_PCT = '34' WALK_PERCENTAGE = BB_PCT K_PCT = '35' STRIKE_PERCENTAGE = K_PCT BB_K = '36' WALKS_PER_STRIKOUT = BB_K OBP = '37' ON_BASE_PERCENTAGE = OBP SLG = '38' SLUGGING = SLG OPS = '39' ON_BASE_PLUS_SLUGGING = OPS ISO = '40' ISOLATED_POWER = ISO BABIP = '41' BATTING_AVERAGE_FOR_BALLS_IN_PLAY = BABIP GB_FB = '42' GROUND_BALLS_PER_FLY_BALL = GB_FB LD_PCT = '43' GB_PCT = '44' GROUND_BALL_PERCENTAGE = GB_PCT FB_PCT = '45' FLY_BALL_PERCENTAGE = FB_PCT IFFB_PCT = '46' HR_FB = '47' HOME_RUNS_PER_FLY_BALL = HR_FB IFH_PCT = '48' INFIELD_HIT_PERCENTAGE = IFH_PCT BUH_PCT = '49' WOBA = '50' WEIGHTED_ON_BASE_AVERAGE = WOBA WRAA = '51' WRC = '52' WEIGHTED_RUNS_CREATED = WRC BATTING = '53' BAT = BATTING FIELDING = '54' FLD = FIELDING REPLACEMENT = '55' REP = REPLACEMENT POSITIONAL = '56' POS = POSITIONAL RAR = '57' RUNS_ABOVE_REPLACEMENT = RAR WAR = '58' WINS_ABOVE_REPLACEMENT = WAR DOLLARS = '59' DOL = DOLLARS DOLLARS_VALUE = DOLLARS SPD = '60' WRC_PLUS = '61' WEIGHTED_RUNS_CREATED_PLUS = WRC_PLUS WPA = '62' WIN_PROBABILITY_ADDED = WPA NEGATIVE_WPA = '63' NEGATIVE_WIN_PROBABILITY_ADDED = NEGATIVE_WPA POSITIVE_WPA = '64' POSITIVE_WIN_PROBABILITY_ADDED = POSITIVE_WPA RE24 = '65' REW = '66' PLI = '67' PHLI = '68' PH = '69' WPA_LI = '70' CLUTCH = '71' FB_PCT_PITCH = '72' FB_PCT_2 = FB_PCT_PITCH FBV = '73' SL_PCT = '74' SLV = '75' CT_PCT = '76' CTV = '77' CB_PCT = '78' CBV = '79' CH_PCT = '80' CHV = '81' SF_PCT = '82' SFV = '83' KN_PCT = '84' KNV = '85' XX_PCT = '86' PO_PCT = '87' WFB = '88' WSL = '89' WCT = '90' WCB = '91' WCH = '92' WSF = '93' WKN = '94' WFB_C = '95' WSL_C = '96' WCT_C = '97' WCB_C = '98' WCH_C = '99' WSF_C = '100' WKN_C = '101' O_SWING_PCT = '102' OSWING_PCT = O_SWING_PCT Z_SWING_PCT = '103' ZSWING_PCT = Z_SWING_PCT SWING_PCT = '104' O_CONTACT_PCT = '105' OCONTACT_PCT = O_CONTACT_PCT Z_CONTACT_PCT = '106' ZCONTACT_PCT = Z_CONTACT_PCT CONTACT_PCT = '107' ZONE_PCT = '108' F_STRIKE_PCT = '109' FSTRIKE_PCT = F_STRIKE_PCT SWSTR_PCT = '110' BASE_RUNNING = '111' BSR = BASE_RUNNING FA_PCT_SC = '112' FT_PCT_SC = '113' FC_PCT_SC = '114' FS_PCT_SC = '115' FO_PCT_SC = '116' SI_PCT_SC = '117' SL_PCT_SC = '118' CU_PCT_SC = '119' KC_PCT_SC = '120' EP_PCT_SC = '121' CH_PCT_SC = '122' SC_PCT_SC = '123' KN_PCT_SC = '124' UN_PCT_SC = '125' VFA_SC = '126' VFT_SC = '127' VFC_SC = '128' VFS_SC = '129' VFO_SC = '130' VSI_SC = '131' VSL_SC = '132' VCU_SC = '133' VKC_SC = '134' VEP_SC = '135' VCH_SC = '136' VSC_SC = '137' VKN_SC = '138' FA_X_SC = '139' FT_X_SC = '140' FC_X_SC = '141' FS_X_SC = '142' FO_X_SC = '143' SI_X_SC = '144' SL_X_SC = '145' CU_X_SC = '146' KC_X_SC = '147' EP_X_SC = '148' CH_X_SC = '149' SC_X_SC = '150' KN_X_SC = '151' FA_Z_SC = '152' FT_Z_SC = '153' FC_Z_SC = '154' FS_Z_SC = '155' FO_Z_SC = '156' SI_Z_SC = '157' SL_Z_SC = '158' CU_Z_SC = '159' KC_Z_SC = '160' EP_Z_SC = '161' CH_Z_SC = '162' SC_Z_SC = '163' KN_Z_SC = '164' WFA_SC = '165' WFT_SC = '166' WFC_SC = '167' WFS_SC = '168' WFO_SC = '169' WSI_SC = '170' WSL_SC = '171' WCU_SC = '172' WKC_SC = '173' WEP_SC = '174' WCH_SC = '175' WSC_SC = '176' WKN_SC = '177' WFA_C_SC = '178' WFT_C_SC = '179' WFC_C_SC = '180' WFS_C_SC = '181' WFO_C_SC = '182' WSI_C_SC = '183' WSL_C_SC = '184' WCU_C_SC = '185' WKC_C_SC = '186' WEP_C_SC = '187' WCH_C_SC = '188' WSC_C_SC = '189' WKN_C_SC = '190' O_SWING_PCT_SC = '191' Z_SWING_PCT_SC = '192' SWING_PCT_SC = '193' O_CONTACT_PCT_SC = '194' Z_CONTACT_PCT_SC = '195' CONTACT_PCT_SC = '196' ZONE_PCT_SC = '197' PACE = '198' DEFENSE = '199' DEF = DEFENSE WSB = '200' UBR = '201' AGE_RNG = '202' OFFENSE = '203' OFF = OFFENSE LEAGUE = '204' LG = LEAGUE WGDP = '205' wGDP = WGDP PULL_PCT = '206' CENT_PCT = '207' OPPO_PCT = '208' SOFT_PCT = '209' MED_PCT = '210' HARD_PCT = '211' TTO_PCT = '212' CH_PCT_PI = '213' CS_PCT_PI = '214' CU_PCT_PI = '215' FA_PCT_PI = '216' FC_PCT_PI = '217' FS_PCT_PI = '218' KN_PCT_PI = '219' SB_PCT_PI = '220' SI_PCT_PI = '221' SL_PCT_PI = '222' XX_PCT_PI = '223' VCH_PI = '224' VCS_PI = '225' VCU_PI = '226' VFA_PI = '227' VFC_PI = '228' VFS_PI = '229' VKN_PI = '230' VSB_PI = '231' VSI_PI = '232' VSL_PI = '233' VXX_PI = '234' CH_X_PI = '235' CS_X_PI = '236' CU_X_PI = '237' FA_X_PI = '238' FC_X_PI = '239' FS_X_PI = '240' KN_X_PI = '241' SB_X_PI = '242' SI_X_PI = '243' SL_X_PI = '244' XX_X_PI = '245' CH_Z_PI = '246' CS_Z_PI = '247' CU_Z_PI = '248' FA_Z_PI = '249' FC_Z_PI = '250' FS_Z_PI = '251' KN_Z_PI = '252' SB_Z_PI = '253' SI_Z_PI = '254' SL_Z_PI = '255' XX_Z_PI = '256' WCH_PI = '257' WCS_PI = '258' WCU_PI = '259' WFA_PI = '260' WFC_PI = '261' WFS_PI = '262' WKN_PI = '263' WSB_PI = '264' WSI_PI = '265' WSL_PI = '266' WXX_PI = '267' WCH_C_PI = '268' WCS_C_PI = '269' WCU_C_PI = '270' WFA_C_PI = '271' WFC_C_PI = '272' WFS_C_PI = '273' WKN_C_PI = '274' WSB_C_PI = '275' WSI_C_PI = '276' WSL_C_PI = '277' WXX_C_PI = '278' O_SWING_PCT_PI = '279' OSWING_PCT_PI = O_SWING_PCT_PI Z_SWING_PCT_PI = '280' ZSWING_PCT_PI = Z_SWING_PCT_PI SWING_PCT_PI = '281' O_CONTACT_PCT_PI = '282' OCONTACT_PCT_PI = O_CONTACT_PCT_PI Z_CONTACT_PCT_PI = '283' ZCONTACT_PCT_PI = Z_CONTACT_PCT_PI CONTACT_PCT_PI = '284' ZONE_PCT_PI = '285' PACE_PI = '286' FRAMING = '287' FRM = FRAMING AVG_PLUS = '288' BB_PCT_PLUS = '289' K_PCT_PLUS = '290' OBP_PLUS = '291' SLG_PLUS = '292' ISO_PLUS = '293' BABIP_PLUS = '294' LD_PCT_PLUS = '295' LD_PLUS_PCT = LD_PCT_PLUS GB_PCT_PLUS = '296' FB_PCT_PLUS = '297' HR_FB_PCT_PLUS = '298' PULL_PCT_PLUS = '299' CENT_PCT_PLUS = '300' OPPO_PCT_PLUS = '301' SOFT_PCT_PLUS = '302' MED_PCT_PLUS = '303' HARD_PCT_PLUS = '304' EV = '305' LA = '306' BARRELS = '307' BARREL_PCT = '308' MAXEV = '309' HARDHIT = '310' HARDHIT_PCT = '311' EVENTS = '312' CSTR_PCT = '313' CSW_PCT = '314' XBA = '315' XSLG = '316' XWOBA = '317' LEGACY_WAR = '318'
class Normalize(object): def __init__(self, mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5), inplace=False): self.mean = mean self.std = std self.inplace = inplace def __call__(self, x): if (isinstance(x, torch.Tensor) and (len(x.shape) == 3)): x = F.normalize(x, self.mean, self.std, self.inplace) return x def __repr__(self): return (self.__class__.__name__ + '()')
class DepthWise(Module): def __init__(self, in_c, out_c, residual=False, kernel=(3, 3), stride=(2, 2), padding=(1, 1), groups=1): super(DepthWise, self).__init__() self.residual = residual self.layers = nn.Sequential(ConvBlock(in_c, out_c=groups, kernel=(1, 1), padding=(0, 0), stride=(1, 1)), ConvBlock(groups, groups, groups=groups, kernel=kernel, padding=padding, stride=stride), LinearBlock(groups, out_c, kernel=(1, 1), padding=(0, 0), stride=(1, 1))) def forward(self, x): short_cut = None if self.residual: short_cut = x x = self.layers(x) if self.residual: output = (short_cut + x) else: output = x return output
class ModelTest(unittest.TestCase): def check_parameter_count(self, model, target_in_m): count = (model.count_params() / (10 ** 6)) msg = '{} params #{}M suppose to be #{}M.'.format(model.name, count, target_in_m) self.assertAlmostEqual(target_in_m, count, msg=msg, delta=0.1) def check_penultimate_shape(self, model, target_shape): layer = model.get_layer('last_relu') if (K.image_data_format() == 'channels_first'): shape = layer.input_shape[2:] else: shape = layer.input_shape[1:3] self.assertEqual(shape, target_shape) def test_cifar_10(self): model = nasnet.cifar10() self.check_parameter_count(model, 3.3) self.check_penultimate_shape(model, (8, 8)) aux_model = nasnet.cifar10(aux_output=True) self.check_parameter_count(aux_model, 4.9) self.assertEqual(len(aux_model.output), 2) def test_mobile(self): model = nasnet.mobile() self.check_parameter_count(model, 5.3) self.check_penultimate_shape(model, (7, 7)) aux_model = nasnet.mobile(aux_output=True) self.check_parameter_count(aux_model, 7.7) self.assertEqual(len(aux_model.output), 2) def test_large(self): model = nasnet.large() self.check_parameter_count(model, 88.9) self.check_penultimate_shape(model, (11, 11)) aux_model = nasnet.large(aux_output=True) self.check_parameter_count(aux_model, 93.5) self.assertEqual(len(aux_model.output), 2) def test_channel_first(self): K.set_image_data_format('channels_first') model = nasnet.cifar10() self.check_parameter_count(model, 3.3) self.check_penultimate_shape(model, (8, 8)) K.set_image_data_format('channels_last')
class ConfigManger(): DEFAULT_CONFIG = '' def __init__(self, DEFAULT_CONFIG_PATH: str=None, verbose=True, integrality_check=True) -> None: self.parsed_args: Dict[(str, Any)] = YAMLArgParser(verbose=verbose) if (DEFAULT_CONFIG_PATH is None): warnings.warn('No default yaml is provided, only used for parser input arguments.', UserWarning) return self.SET_DEFAULT_CONFIG_PATH(DEFAULT_CONFIG_PATH) if self.parsed_args.get('Config'): if Path(self.parsed_args['Config']).is_dir(): self.parsed_args['Config'] = os.path.join(self.parsed_args['Config'], 'config.yaml') self.default_config: Dict[(str, Any)] = yaml_load(self.parsed_args.get('Config', self.DEFAULT_CONFIG), verbose=verbose) self.merged_config: Dict[(str, Any)] = dict_merge(self.default_config, self.parsed_args) if integrality_check: self._check_integrality(self.merged_config) if verbose: print('Merged args:') pprint(self.merged_config) def SET_DEFAULT_CONFIG_PATH(cls, default_config_path: str) -> None: path: Path = Path(default_config_path) assert path.exists(), path assert path.is_file(), path assert (path.with_suffix('.yaml') or path.with_suffix('.yml')) cls.DEFAULT_CONFIG = str(default_config_path) def _check_integrality(merged_dict=Dict[(str, Any)]): assert merged_dict.get('Arch'), f'Merged dict integrity check failed,{merged_dict.keys()}' assert merged_dict.get('Optim'), f'Merged dict integrity check failed,{merged_dict.keys()}' assert merged_dict.get('Scheduler'), f'Merged dict integrity check failed,{merged_dict.keys()}' assert merged_dict.get('Trainer'), f'Merged dict integrity check failed,{merged_dict.keys()}' def config(self): try: config = self.merged_config except AttributeError: config = self.parsed_args from collections import defaultdict return defaultdict((lambda : None), config)
class BLS2017Model(nn.Module): def __init__(self, num_filters=192): super(BLS2017Model, self).__init__() self.conv1 = nn.Conv2d(3, num_filters, 9, stride=4, padding=4) self.gdn1 = gdn.GDN(num_filters) self.conv2 = nn.Conv2d(num_filters, num_filters, 5, stride=2, padding=2) self.gdn2 = gdn.GDN(num_filters) self.conv3 = nn.Conv2d(num_filters, num_filters, 5, stride=2, padding=2) self.entropy_bottleneck = entropy_model.EntropyBottleneck(num_filters) self.deconv1 = nn.ConvTranspose2d(num_filters, num_filters, 5, stride=2, padding=2, output_padding=1) self.igdn2 = gdn.GDN(num_filters, inverse=True) self.deconv2 = nn.ConvTranspose2d(num_filters, num_filters, 5, stride=2, padding=2, output_padding=1) self.igdn3 = gdn.GDN(num_filters, inverse=True) self.deconv3 = nn.ConvTranspose2d(num_filters, 3, 9, stride=4, padding=4, output_padding=3) def encode(self, x): x = self.conv1(x) x = self.gdn1(x) x = self.conv2(x) x = self.gdn2(x) x = self.conv3(x) return x def decode(self, x): x = self.deconv1(x) x = self.igdn2(x) x = self.deconv2(x) x = self.igdn3(x) x = self.deconv3(x) return x def forward(self, x): y = self.encode(x) (y_tilde, likelihoods) = self.entropy_bottleneck(y) x_tilde = self.decode(y_tilde) return (x_tilde, likelihoods)
def where(condition, input, other): if is_encrypted_tensor(condition): return ((condition * input) + ((1 - condition) * other)) elif torch.is_tensor(condition): condition = condition.float() return ((input * condition) + (other * (1 - condition)))
def mIoU_parser(): parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter) parser.add_argument('--dataset', type=str, default=(basedir + '/Dataset'), help='path to dataset') parser.add_argument('--set_name', type=str, default='val.txt', help='name for set') parser.add_argument('--GT_dir_name', type=str, default='SegmentationClass', help='name for ground truth directory') parser.add_argument('--Pred_dir_name', type=str, default='CRF_masks', help='name for prediction directory') parser.add_argument('--classes', type=int, default=21, help='number of classes') args = parser.parse_args() print('{:{}}: {}'.format('Dataset', SPACE, args.dataset)) print('{:{}}: {}'.format('Set name', SPACE, args.set_name)) print('{:{}}: {}'.format('ground truth directory name', SPACE, args.GT_dir_name)) print('{:{}}: {}'.format('Prediction directory name', SPACE, args.Pred_dir_name)) print('{:{}}: {}'.format('Classes', SPACE, args.classes)) if (not os.path.isdir(args.dataset)): parser.error('Wrong dataset path') if (not os.path.isfile(((args.dataset + '/') + args.set_name))): parser.error('Wrong set name') if (not os.path.isdir(((args.dataset + '/') + args.GT_dir_name))): parser.error('Wrong ground truth directory name') if (not os.path.isdir(((args.dataset + '/') + args.Pred_dir_name))): parser.error('Wrong prediction directory name') return args
class Merge_Run(nn.Module): def __init__(self, in_channels, out_channels, init='xavier', ksize=3, stride=1, pad=1, dilation=1): super(Merge_Run, self).__init__() self.body1 = nn.Sequential(nn.Conv2d(in_channels, out_channels, ksize, stride, pad), nn.LeakyReLU(negative_slope=0.2, inplace=True)) self.body2 = nn.Sequential(nn.Conv2d(in_channels, out_channels, ksize, stride, 2, 2), nn.LeakyReLU(negative_slope=0.2, inplace=True)) self.body3 = nn.Sequential(nn.Conv2d((in_channels * 2), out_channels, ksize, stride, pad), nn.LeakyReLU(negative_slope=0.2, inplace=True)) def forward(self, x): out1 = self.body1(x) out2 = self.body2(x) c = torch.cat([out1, out2], dim=1) c_out = self.body3(c) out = (c_out + x) return out
.parametrize('emitter_type', ['GradientArborescenceEmitter', 'EvolutionStrategyEmitter'], ids=['GAEmitter', 'ESEmitter']) .parametrize('wrong_array,offsets', [('solution_batch', [(0, 1), (1, 0)]), ('objective_batch', [(1,)]), ('measures_batch', [(0, 1), (1, 0)]), ('jacobian_batch', [(0, 0, 1), (0, 1, 0), (1, 0, 0)]), ('status_batch', [(1,)]), ('value_batch', [(1,)])]) def test_tell_arguments_incorrect_shape(emitter_type, wrong_array, offsets): batch_size = 3 archive = GridArchive(solution_dim=1, dims=[10], ranges=[((- 1.0), 1.0)]) if (emitter_type == 'GradientArborescenceEmitter'): emitter = GradientArborescenceEmitter(archive, x0=np.array([0]), sigma0=1.0, lr=0.1, batch_size=batch_size) elif (emitter_type == 'EvolutionStrategyEmitter'): emitter = EvolutionStrategyEmitter(archive, x0=np.array([0]), sigma0=1.0, ranker='imp', batch_size=batch_size) solution_batch = np.ones((batch_size, archive.solution_dim)) objective_batch = np.ones(batch_size) measures_batch = np.ones((batch_size, archive.measure_dim)) jacobian_batch = np.ones((batch_size, (archive.measure_dim + 1), archive.solution_dim)) status_batch = np.ones(batch_size) value_batch = np.ones(batch_size) for offset in offsets: if (wrong_array == 'solution_batch'): solution_batch = np.ones(((batch_size + offset[0]), (archive.solution_dim + offset[1]))) elif (wrong_array == 'objective_batch'): objective_batch = np.ones((batch_size + offset[0])) elif (wrong_array == 'measures_batch'): measures_batch = np.ones(((batch_size + offset[0]), (archive.measure_dim + offset[1]))) elif (wrong_array == 'jacobian_batch'): jacobian_batch = np.ones(((batch_size + offset[0]), ((archive.measure_dim + 1) + offset[1]), (archive.solution_dim + offset[2]))) elif (wrong_array == 'status_batch'): status_batch = np.ones((batch_size + offset[0])) elif (wrong_array == 'value_batch'): value_batch = np.ones((batch_size + offset[0])) if isinstance(emitter, GradientArborescenceEmitter): with pytest.raises(ValueError): emitter.tell_dqd(solution_batch, objective_batch, measures_batch, jacobian_batch, {'status': status_batch, 'value': value_batch}) if (wrong_array == 'jacobian_batch'): return with pytest.raises(ValueError): emitter.tell(solution_batch, objective_batch, measures_batch, {'status': status_batch, 'value': value_batch})
def attention(x): params = dict(activation='relu', padding='valid', kernel_regularizer=l2(1e-05)) x = Conv2D(8, kernel_size=3, **params)(x) x = Conv2D(16, kernel_size=3, **params)(x) x = Conv2D(32, kernel_size=3, **params)(x) x = Conv2D(1, kernel_size=3)(x) x = MaxPooling2D(pool_size=8)(x) x = SampleSoftmax(squeeze_channels=True, smooth=0.0001)(x) return x
def define_net_r(opt): network_type = opt.pop('type') net_r = dynamic_instantiation(_arch_modules, network_type, opt) return net_r
def get_charset(lang): global _CHARSETS cls_or_obj = _CHARSETS[lang] if isinstance(cls_or_obj, type): _CHARSETS[lang] = cls_or_obj() return _CHARSETS[lang]
class GeneralDataset(Dataset): def __init__(self, root, transform=None, target_transform=None, top_k=(1, 5), keep_rgb=False): self.data_set = datasets.ImageFolder(root) self.classes = self.data_set.classes self.root = root self.transform = transform self.target_transform = target_transform self.keep_rgb = keep_rgb self._update_evaluator(top_k) def __getitem__(self, index: int): (image, target) = self.data_set.__getitem__(index) image = default_converter(image, rgb=self.keep_rgb) if (self.transform is not None): image = self.transform(image) if (self.target_transform is not None): target = self.target_transform(target) return (image, target) def __len__(self) -> int: return self.data_set.__len__() def _update_evaluator(self, top_k): self.evaluator = GeneralEvaluator(self.classes, top_k=top_k) def get_classes(self): return self.classes def __repr__(self): return (((self.__class__.__name__ + ' (') + self.root) + ')')
def get_and_print_layers_to_use_halut(model: torch.nn.Module) -> list[str]: all_layers = [] def layers(module: torch.nn.Module, prefix: str='') -> None: if isinstance(module, (HalutLinear, HalutConv2d)): all_layers.append(prefix[:(- 1)]) for (name, child) in module._modules.items(): if (child is not None): layers(child, ((prefix + name) + '.')) layers(model) del layers print(all_layers) return all_layers
def test_shape_validation_during_creation(): tensor = torch.tensor(np.random.rand(3)) with pytest.raises(ValueError): box_tensor = MinDeltaBoxTensor(tensor) tensor = torch.tensor(np.random.rand(3, 11)) with pytest.raises(ValueError): box_tensor = MinDeltaBoxTensor(tensor) tensor = torch.tensor(np.random.rand(3, 3, 3)) with pytest.raises(ValueError): box_tensor = MinDeltaBoxTensor(tensor)
def get_d_UB(l, u, func, dfunc): diff = (lambda d, l: (((func(d) - func(l)) / (d - l)) - dfunc(d))) max_iter = 1000 ub = (- l) d = (ub / 2) device = l.device lb = torch.zeros(l.shape, device=device) keep_search = torch.ones(l.shape, device=device).byte() for i in range(max_iter): t = diff(d[keep_search], l[keep_search]) idx = ((t < 0) + (t.abs() > 0.01)) keep_search[keep_search] = (idx > 0) if (keep_search.sum() == 0): break t = t[(idx > 0)] idx = (t > 0) keep_search_copy = keep_search.data.clone() keep_search_copy[keep_search_copy] = idx ub[keep_search_copy] = d[keep_search_copy] d[keep_search_copy] = ((d[keep_search_copy] + lb[keep_search_copy]) / 2) idx = (t < 0) keep_search_copy = keep_search.data.clone() keep_search_copy[keep_search_copy] = idx lb[keep_search_copy] = d[keep_search_copy] d[keep_search_copy] = ((d[keep_search_copy] + ub[keep_search_copy]) / 2) return d
class _ClassInfo(_BlockInfo): def __init__(self, name, class_or_struct, clean_lines, linenum): _BlockInfo.__init__(self, False) self.name = name self.starting_linenum = linenum self.is_derived = False if (class_or_struct == 'struct'): self.access = 'public' self.is_struct = True else: self.access = 'private' self.is_struct = False initial_indent = Match('^( *)\\S', clean_lines.raw_lines[linenum]) if initial_indent: self.class_indent = len(initial_indent.group(1)) else: self.class_indent = 0 self.last_line = 0 depth = 0 for i in range(linenum, clean_lines.NumLines()): line = clean_lines.elided[i] depth += (line.count('{') - line.count('}')) if (not depth): self.last_line = i break def CheckBegin(self, filename, clean_lines, linenum, error): if Search('(^|[^:]):($|[^:])', clean_lines.elided[linenum]): self.is_derived = True def CheckEnd(self, filename, clean_lines, linenum, error): indent = Match('^( *)\\}', clean_lines.elided[linenum]) if (indent and (len(indent.group(1)) != self.class_indent)): if self.is_struct: parent = ('struct ' + self.name) else: parent = ('class ' + self.name) error(filename, linenum, 'whitespace/indent', 3, ('Closing brace should be aligned with beginning of %s' % parent))
def area(x, y): ymax = np.max(y) xmax = np.max(x) bin_mask = np.zeros((ymax, xmax)) (rr, cc) = polygon(y, x) bin_mask[(rr, cc)] = 1 area = np.sum(bin_mask) return area
class CIFAR100(DATASET): _target_: str = 'dataset_loaders.load_CIFAR100' name: str = 'CIFAR100' IN_CHANNEL: int = 3 N_CLASSES: int = 100 IMG_SIZE: Tuple[int] = field(default_factory=(lambda : (32, 32)))
def mkdir(directory: str) -> None: return pathlib.Path(directory).mkdir(parents=True, exist_ok=True)
class ROSDataManagerConfig(base_datamanager.VanillaDataManagerConfig): _target: Type = field(default_factory=(lambda : ROSDataManager)) dataparser: ROSDataParserConfig = ROSDataParserConfig() publish_training_posearray: bool = True data_update_freq: float = 5.0 num_training_images: int = 500
class CheckpointFunction(torch.autograd.Function): def forward(ctx, run_function, preserve_rng_state, *args): check_backward_validity(args) ctx.run_function = run_function ctx.preserve_rng_state = preserve_rng_state ctx.fwd_cuda_rng_state_tracker = get_cuda_rng_tracker().get_states() if preserve_rng_state: ctx.fwd_cpu_state = torch.get_rng_state() ctx.had_cuda_in_fwd = False if torch.cuda._initialized: ctx.had_cuda_in_fwd = True (ctx.fwd_gpu_devices, ctx.fwd_gpu_states) = get_device_states(*args) ctx.save_for_backward(*args) with torch.no_grad(): outputs = run_function(*args) return outputs def backward(ctx, *args): if (not torch.autograd._is_checkpoint_valid()): raise RuntimeError('Checkpointing is not compatible with .grad(), please use .backward() if possible') inputs = ctx.saved_tensors rng_devices = [] if (ctx.preserve_rng_state and ctx.had_cuda_in_fwd): rng_devices = ctx.fwd_gpu_devices bwd_cuda_rng_state_tracker = get_cuda_rng_tracker().get_states() with torch.random.fork_rng(devices=rng_devices, enabled=ctx.preserve_rng_state): if ctx.preserve_rng_state: torch.set_rng_state(ctx.fwd_cpu_state) if ctx.had_cuda_in_fwd: set_device_states(ctx.fwd_gpu_devices, ctx.fwd_gpu_states) get_cuda_rng_tracker().set_states(ctx.fwd_cuda_rng_state_tracker) detached_inputs = detach_variable(inputs) with torch.enable_grad(): outputs = ctx.run_function(*detached_inputs) if isinstance(outputs, torch.Tensor): outputs = (outputs,) get_cuda_rng_tracker().set_states(bwd_cuda_rng_state_tracker) torch.autograd.backward(outputs, args) grads = tuple(((inp.grad if isinstance(inp, torch.Tensor) else inp) for inp in detached_inputs)) return ((None, None) + grads)
def render_batch(npy_dir, execute_python='./scripts/visualize_motion.sh', mode='sequence'): os.system(f'{execute_python} {npy_dir} {mode}')
def test_components_vs_sklearn(): def check_components(Estimator, n_components, shape): X = DATA[shape] pca = Estimator(n_components, **KWDS[Estimator]).fit(X) skpca = SKPCA(n_components).fit(X) assert_columns_allclose_upto_sign(pca.components_.T, skpca.components_.T) for Estimator in ESTIMATORS: for shape in SHAPES: for n_components in N_COMPONENTS: (yield (check_components, Estimator, n_components, shape))
def base_cli_dir_args(image: pathlib.Path, mask: pathlib.Path) -> typing.List[str]: return f'{image.parent} -m {mask.parent}'.split()
class DistogramHead(nn.Module): def __init__(self, c_z, no_bins, **kwargs): super(DistogramHead, self).__init__() self.c_z = c_z self.no_bins = no_bins self.linear = Linear(self.c_z, self.no_bins, init='final') def forward(self, z): logits = self.linear(z) logits = (logits + logits.transpose((- 2), (- 3))) return logits
def _config_validation(config): if (config == None): return None if (isinstance(config, dict) != True): with open(config, 'r') as conf_file: import yaml config = yaml.safe_load(conf_file) from schema import Schema conf_schema = Schema({'pattern_switch': Schema({str: bool}, error='The format of the pattern config is wrong.')}) return conf_schema.validate(config)
class ObjectData(): def __init__(self, id, x, y): self.id = id self.x = x self.y = y self.type = 'other'
class SUBDATA(data.Dataset): def __init__(self): self.data = np.load('./data/sub_data_training.npy', allow_pickle=True) def __getitem__(self, index): return (self.data[index][0], self.data[index][1], self.data[index][2], self.data[index][3], self.data[index][4], self.data[index][5], self.data[index][6], self.data[index][7]) def __len__(self): return ((len(self.data) // 8) * 8)
def train(args, train_dataset, model, tokenizer, teacher=None): if (args.local_rank in [(- 1), 0]): tb_writer = SummaryWriter() args.train_batch_size = (args.per_gpu_train_batch_size * max(1, args.n_gpu)) train_sampler = (RandomSampler(train_dataset) if (args.local_rank == (- 1)) else DistributedSampler(train_dataset)) train_dataloader = DataLoader(train_dataset, sampler=train_sampler, batch_size=args.train_batch_size) if (args.max_steps > 0): t_total = args.max_steps args.num_train_epochs = ((args.max_steps // (len(train_dataloader) // args.gradient_accumulation_steps)) + 1) else: t_total = ((len(train_dataloader) // args.gradient_accumulation_steps) * args.num_train_epochs) no_decay = ['bias', 'LayerNorm.weight'] optimizer_grouped_parameters = [{'params': [p for (n, p) in model.named_parameters() if (not any(((nd in n) for nd in no_decay)))], 'weight_decay': args.weight_decay}, {'params': [p for (n, p) in model.named_parameters() if any(((nd in n) for nd in no_decay))], 'weight_decay': 0.0}] optimizer = AdamW(optimizer_grouped_parameters, lr=args.learning_rate, eps=args.adam_epsilon) scheduler = get_linear_schedule_with_warmup(optimizer, num_warmup_steps=args.warmup_steps, num_training_steps=t_total) if (os.path.isfile(os.path.join(args.model_name_or_path, 'optimizer.pt')) and os.path.isfile(os.path.join(args.model_name_or_path, 'scheduler.pt'))): optimizer.load_state_dict(torch.load(os.path.join(args.model_name_or_path, 'optimizer.pt'))) scheduler.load_state_dict(torch.load(os.path.join(args.model_name_or_path, 'scheduler.pt'))) if args.fp16: try: from apex import amp except ImportError: raise ImportError('Please install apex from to use fp16 training.') (model, optimizer) = amp.initialize(model, optimizer, opt_level=args.fp16_opt_level) if (args.n_gpu > 1): model = nn.DataParallel(model) if (args.local_rank != (- 1)): model = nn.parallel.DistributedDataParallel(model, device_ids=[args.local_rank], output_device=args.local_rank, find_unused_parameters=True) logger.info('***** Running training *****') logger.info(' Num examples = %d', len(train_dataset)) logger.info(' Num Epochs = %d', args.num_train_epochs) logger.info(' Instantaneous batch size per GPU = %d', args.per_gpu_train_batch_size) logger.info(' Total train batch size (w. parallel, distributed & accumulation) = %d', ((args.train_batch_size * args.gradient_accumulation_steps) * (torch.distributed.get_world_size() if (args.local_rank != (- 1)) else 1))) logger.info(' Gradient Accumulation steps = %d', args.gradient_accumulation_steps) logger.info(' Total optimization steps = %d', t_total) global_step = 1 epochs_trained = 0 steps_trained_in_current_epoch = 0 if os.path.exists(args.model_name_or_path): try: checkpoint_suffix = args.model_name_or_path.split('-')[(- 1)].split('/')[0] global_step = int(checkpoint_suffix) epochs_trained = (global_step // (len(train_dataloader) // args.gradient_accumulation_steps)) steps_trained_in_current_epoch = (global_step % (len(train_dataloader) // args.gradient_accumulation_steps)) logger.info(' Continuing training from checkpoint, will skip to saved global_step') logger.info(' Continuing training from epoch %d', epochs_trained) logger.info(' Continuing training from global step %d', global_step) logger.info(' Will skip the first %d steps in the first epoch', steps_trained_in_current_epoch) except ValueError: logger.info(' Starting fine-tuning.') (tr_loss, logging_loss) = (0.0, 0.0) model.zero_grad() train_iterator = trange(epochs_trained, int(args.num_train_epochs), desc='Epoch', disable=(args.local_rank not in [(- 1), 0])) set_seed(args) for _ in train_iterator: epoch_iterator = tqdm(train_dataloader, desc='Iteration', disable=(args.local_rank not in [(- 1), 0])) for (step, batch) in enumerate(epoch_iterator): if (steps_trained_in_current_epoch > 0): steps_trained_in_current_epoch -= 1 continue model.train() if (teacher is not None): teacher.eval() batch = tuple((t.to(args.device) for t in batch)) inputs = {'input_ids': batch[0], 'attention_mask': batch[1], 'start_positions': batch[3], 'end_positions': batch[4]} if (args.model_type != 'distilbert'): inputs['token_type_ids'] = (None if (args.model_type == 'xlm') else batch[2]) if (args.model_type in ['xlnet', 'xlm']): inputs.update({'cls_index': batch[5], 'p_mask': batch[6]}) if args.version_2_with_negative: inputs.update({'is_impossible': batch[7]}) outputs = model(**inputs) (loss, start_logits_stu, end_logits_stu) = outputs if (teacher is not None): if ('token_type_ids' not in inputs): inputs['token_type_ids'] = (None if (args.teacher_type == 'xlm') else batch[2]) with torch.no_grad(): (start_logits_tea, end_logits_tea) = teacher(input_ids=inputs['input_ids'], token_type_ids=inputs['token_type_ids'], attention_mask=inputs['attention_mask']) assert (start_logits_tea.size() == start_logits_stu.size()) assert (end_logits_tea.size() == end_logits_stu.size()) loss_fct = nn.KLDivLoss(reduction='batchmean') loss_start = (loss_fct(nn.functional.log_softmax((start_logits_stu / args.temperature), dim=(- 1)), nn.functional.softmax((start_logits_tea / args.temperature), dim=(- 1))) * (args.temperature ** 2)) loss_end = (loss_fct(nn.functional.log_softmax((end_logits_stu / args.temperature), dim=(- 1)), nn.functional.softmax((end_logits_tea / args.temperature), dim=(- 1))) * (args.temperature ** 2)) loss_ce = ((loss_start + loss_end) / 2.0) loss = ((args.alpha_ce * loss_ce) + (args.alpha_squad * loss)) if (args.n_gpu > 1): loss = loss.mean() if (args.gradient_accumulation_steps > 1): loss = (loss / args.gradient_accumulation_steps) if args.fp16: with amp.scale_loss(loss, optimizer) as scaled_loss: scaled_loss.backward() else: loss.backward() tr_loss += loss.item() if (((step + 1) % args.gradient_accumulation_steps) == 0): if args.fp16: nn.utils.clip_grad_norm_(amp.master_params(optimizer), args.max_grad_norm) else: nn.utils.clip_grad_norm_(model.parameters(), args.max_grad_norm) optimizer.step() scheduler.step() model.zero_grad() global_step += 1 if ((args.local_rank in [(- 1), 0]) and (args.logging_steps > 0) and ((global_step % args.logging_steps) == 0)): if ((args.local_rank == (- 1)) and args.evaluate_during_training): results = evaluate(args, model, tokenizer) for (key, value) in results.items(): tb_writer.add_scalar('eval_{}'.format(key), value, global_step) tb_writer.add_scalar('lr', scheduler.get_lr()[0], global_step) tb_writer.add_scalar('loss', ((tr_loss - logging_loss) / args.logging_steps), global_step) logging_loss = tr_loss if ((args.local_rank in [(- 1), 0]) and (args.save_steps > 0) and ((global_step % args.save_steps) == 0)): output_dir = os.path.join(args.output_dir, 'checkpoint-{}'.format(global_step)) if (not os.path.exists(output_dir)): os.makedirs(output_dir) model_to_save = (model.module if hasattr(model, 'module') else model) model_to_save.save_pretrained(output_dir) tokenizer.save_pretrained(output_dir) torch.save(args, os.path.join(output_dir, 'training_args.bin')) logger.info('Saving model checkpoint to %s', output_dir) torch.save(optimizer.state_dict(), os.path.join(output_dir, 'optimizer.pt')) torch.save(scheduler.state_dict(), os.path.join(output_dir, 'scheduler.pt')) logger.info('Saving optimizer and scheduler states to %s', output_dir) if ((args.max_steps > 0) and (global_step > args.max_steps)): epoch_iterator.close() break if ((args.max_steps > 0) and (global_step > args.max_steps)): train_iterator.close() break if (args.local_rank in [(- 1), 0]): tb_writer.close() return (global_step, (tr_loss / global_step))
def adjust_improvedgt_folders(kitti_folder='kitti_download'): path_getter = gp.GetPath() dataset_folder_path = path_getter.get_data_path() gt_path = os.path.join(dataset_folder_path, kitti_folder) gt_path = os.path.join(gt_path, 'Depth_improved') assert os.path.isdir(gt_path), 'Path to data does not exist' folders = dl.DirLister.get_directories(gt_path) folders = dl.DirLister.include_dirs_by_name(folders, 'proj_depth') for f in folders: (ground_path, camera) = os.path.split(f) ground_path = os.path.split(ground_path)[0] ground_path = os.path.split(ground_path)[0] target_path = os.path.join(ground_path, camera, 'data') if (not os.path.isdir(target_path)): os.makedirs(target_path) else: continue for filepath in glob.glob(os.path.join(f, '*')): shutil.move(filepath, target_path) print(target_path) for f in folders: remove_path = os.path.split(f)[0] remove_path = os.path.split(remove_path)[0] print(remove_path) shutil.rmtree(remove_path, ignore_errors=True)
def main(): parser = argparse.ArgumentParser() parser.add_argument('--model', required=True, help='sentencepiece model to use for decoding') parser.add_argument('--input', required=True, help='input file to decode') parser.add_argument('--input_format', choices=['piece', 'id'], default='piece') args = parser.parse_args() sp = spm.SentencePieceProcessor() sp.Load(args.model) if (args.input_format == 'piece'): def decode(input): return ''.join(sp.DecodePieces(input)) elif (args.input_format == 'id'): def decode(input): return ''.join(sp.DecodeIds(input)) else: raise NotImplementedError def tok2int(tok): return (int(tok) if (tok != '<<unk>>') else 0) with open(args.input, 'r', encoding='utf-8') as h: for line in h: if (args.input_format == 'id'): print(decode(list(map(tok2int, line.rstrip().split())))) elif (args.input_format == 'piece'): print(decode(line.rstrip().split()))
class RSSMPosterior(nn.Module): c: Config def __call__(self, prior, obs_inputs): inputs = jnp.concatenate([prior['det_out'], obs_inputs], (- 1)) hl = nn.relu(nn.Dense(self.c.cell_embed_size)(inputs)) hl = nn.relu(nn.Dense(self.c.cell_embed_size)(hl)) mean = nn.Dense(self.c.cell_stoch_size)(hl) stddev = (nn.softplus(nn.Dense(self.c.cell_stoch_size)((hl + 0.54))) + self.c.cell_min_stddev) dist = tfd.MultivariateNormalDiag(mean, stddev) sample = dist.sample(seed=self.make_rng('sample')) return dict(mean=mean, stddev=stddev, sample=sample, det_out=prior['det_out'], det_state=prior['det_state'], output=jnp.concatenate([sample, prior['det_out']], (- 1)))
def get_llm_packages(): llm_packages = [] for (dirpath, _, _) in os.walk(os.path.join(llm_home, 'bigdl')): print(dirpath) package = dirpath.split((llm_home + os.sep))[1].replace(os.sep, '.') if any((fnmatch.fnmatchcase(package, pat=pattern) for pattern in exclude_patterns)): print('excluding', package) else: llm_packages.append(package) print('including', package) return llm_packages
def constrain_norm(grads: P, preconditioned_grads: P, learning_rate: chex.Numeric, norm_constraint: chex.Numeric=0.001) -> P: sq_norm_grads = tree_inner_product(preconditioned_grads, grads) sq_norm_scaled_grads = (sq_norm_grads * (learning_rate ** 2)) sq_norm_scaled_grads = utils.distribute.pmean_if_pmap(sq_norm_scaled_grads) norm_scale_factor = jnp.sqrt((norm_constraint / sq_norm_scaled_grads)) coefficient = jnp.minimum(norm_scale_factor, 1) constrained_grads = multiply_tree_by_scalar(preconditioned_grads, coefficient) return constrained_grads
.skip(reason='make_bag test needs to be updated') def test_make_bag_regression(): data = synthetic_regression() X_orig = data['full']['X'] y_orig = data['full']['y'] X = np.array(X_orig) y = np.array(y_orig) w = np.ones_like(y, dtype=np.float64) test_size = 0.2 (X_train, X_val, y_train, y_val, w_train, w_val, _, _) = make_bag(X, y, w, test_size=test_size, random_state=1, is_classification=False) num_samples = X.shape[0] num_features = X.shape[1] num_test_expected = ceil((test_size * num_samples)) num_train_expected = (num_samples - num_test_expected) assert (X_train.shape == (num_features, num_train_expected)) assert (X_val.shape == (num_features, num_test_expected)) assert (y_train.shape == (num_train_expected,)) assert (y_val.shape == (num_test_expected,)) assert (w_train.shape == (num_train_expected,)) assert (w_val.shape == (num_test_expected,)) X_all = np.concatenate((X_train.T, X_val.T)) np.array_equal(np.sort(X, axis=0), np.sort(X_all, axis=0))
def get_total(records): record_vals = [item for sublist in records.values() for item in sublist] total_mean_fps = (sum([r['fps_mean'] for r in record_vals]) / len(record_vals)) total_mean_std = (sum([r['fps_std'] for r in record_vals]) / len(record_vals)) return (total_mean_fps, total_mean_std)
def kronecker_product(a, b): siz1 = torch.Size((torch.tensor(a.shape[(- 2):]) * torch.tensor(b.shape[(- 2):]))) res = (a.unsqueeze((- 1)).unsqueeze((- 3)) * b.unsqueeze((- 2)).unsqueeze((- 4))) siz0 = res.shape[:(- 4)] out = res.reshape((siz0 + siz1)) return out
class PairClassifiers(nn.Module): def __init__(self, fdim, num_classes): super().__init__() self.c1 = nn.Linear(fdim, num_classes) self.c2 = nn.Linear(fdim, num_classes) def forward(self, x): z1 = self.c1(x) if (not self.training): return z1 z2 = self.c2(x) return (z1, z2)
class PreActResNet(nn.Module): def __init__(self, block, num_blocks, num_classes=200): super(PreActResNet, self).__init__() self.in_planes = 64 self.conv1 = nn.Conv2d(3, 64, kernel_size=3, stride=1, padding=1, bias=False) self.layer1 = self._make_layer(block, 64, num_blocks[0], stride=1) self.layer2 = self._make_layer(block, 128, num_blocks[1], stride=2) self.layer3 = self._make_layer(block, 256, num_blocks[2], stride=2) self.layer4 = self._make_layer(block, 512, num_blocks[3], stride=2) self.bn = nn.BatchNorm2d((512 * block.expansion)) self.avgpool = nn.AdaptiveAvgPool2d((1, 1)) self.linear = nn.Linear((512 * block.expansion), num_classes) def _make_layer(self, block, planes, num_blocks, stride): strides = ([stride] + ([1] * (num_blocks - 1))) layers = [] for stride in strides: layers.append(block(self.in_planes, planes, stride)) self.in_planes = (planes * block.expansion) return nn.Sequential(*layers) def forward(self, x): out = self.conv1(x) out = self.layer1(out) out = self.layer2(out) out = self.layer3(out) out = self.layer4(out) out = F.relu(self.bn(out)) out = self.avgpool(out) out = out.view(out.size(0), (- 1)) out = self.linear(out) return out
def noam_schedule(step, warmup_step=4000): if (step <= warmup_step): return (step / warmup_step) return ((warmup_step ** 0.5) * (step ** (- 0.5)))
def parse_requirements(fname='requirements.txt', with_version=True): import re import sys from os.path import exists require_fpath = fname def parse_line(line): if line.startswith('-r '): target = line.split(' ')[1] for info in parse_require_file(target): (yield info) else: info = {'line': line} if line.startswith('-e '): info['package'] = line.split('#egg=')[1] elif ('+' in line): info['package'] = line else: pat = (('(' + '|'.join(['>=', '<=', '==', '>', '<'])) + ')') parts = re.split(pat, line, maxsplit=1) parts = [p.strip() for p in parts] info['package'] = parts[0] if (len(parts) > 1): (op, rest) = parts[1:] if (';' in rest): (version, platform_deps) = map(str.strip, rest.split(';')) info['platform_deps'] = platform_deps else: version = rest info['version'] = (op, version) (yield info) def parse_require_file(fpath): with open(fpath, 'r') as f: for line in f.readlines(): line = line.strip() if (line and (not line.startswith('#'))): for info in parse_line(line): (yield info) def gen_packages_items(): if exists(require_fpath): for info in parse_require_file(require_fpath): parts = [info['package']] if (with_version and ('version' in info)): parts.extend(info['version']) if (not sys.version.startswith('3.4')): platform_deps = info.get('platform_deps') if (platform_deps is not None): parts.append((';' + platform_deps)) item = ''.join(parts) (yield item) packages = list(gen_packages_items()) return packages
def mixed_volume(mixture, points, checkin=True): from phcpy.phcpy2c3 import py2c_celcon_initialize_supports as init from phcpy.phcpy2c3 import py2c_celcon_set_type_of_mixture as setmix from phcpy.phcpy2c3 import py2c_celcon_append_lifted_point as applft from phcpy.phcpy2c3 import py2c_celcon_mixed_volume_of_supports as mixvol if checkin: if (not check_mixture(mixture, points)): print('incorrect type of mixture') return (- 1) nbr = len(mixture) init(nbr) setmix(nbr, str(mixture)) for k in range(nbr): for point in points[k]: lpt = list(point) lpt.append(0) applft(len(lpt), (k + 1), str(lpt)) return mixvol()
def _check_model_old_version(model): if hasattr(model.WN[0], 'res_layers'): return True else: return False
def icnr(x, scale=2, init=nn.init.kaiming_normal_): (ni, nf, h, w) = x.shape ni2 = int((ni / (scale ** 2))) k = init(torch.zeros([ni2, nf, h, w])).transpose(0, 1) k = k.contiguous().view(ni2, nf, (- 1)) k = k.repeat(1, 1, (scale ** 2)) k = k.contiguous().view([nf, ni, h, w]).transpose(0, 1) x.data.copy_(k)
def setup_custom_environment(custom_module): if custom_module.endswith('.py'): module = _import_file('detectron2.utils.env.custom_module', custom_module) else: module = importlib.import_module(custom_module) assert (hasattr(module, 'setup_environment') and callable(module.setup_environment)), "Custom environment module defined in {} does not have the required callable attribute 'setup_environment'.".format(custom_module) module.setup_environment()
def test_move_right(board: Board, another_board: Board) -> None: (board, reward) = move_right(board) expected_board = jnp.array([[0, 0, 0, 2], [0, 0, 2, 2], [0, 0, 1, 2], [0, 0, 0, 2]]) assert jnp.array_equal(board, expected_board) assert (reward == 8) (board, reward) = move_right(another_board) expected_board = jnp.array([[0, 0, 2, 2], [2, 3, 1, 2], [2, 3, 1, 2], [0, 0, 2, 3]]) assert jnp.array_equal(board, expected_board) assert (reward == 4)
class Logger(object): def __init__(self, filename): self.terminal = sys.stdout self.log = open(filename, 'a') def write(self, message): self.terminal.write(message) self.log.write(message) def flush(self): pass
def prepare_within_day_indices_for_ground_truth(offset: int) -> np.ndarray: return np.add([1, 2, 3, 6, 9, 12], (11 + offset))
class Metrics(): def __init__(self, residues: dict[(str, float)], isotope_error_range: list[int], cum_mass_threshold: float=0.5, ind_mass_threshold: float=0.1) -> None: self.residues = residues self.isotope_error_range = isotope_error_range self.cum_mass_threshold = cum_mass_threshold self.ind_mass_threshold = ind_mass_threshold def _split_sequences(seq: (list[str] | list[list[str]])) -> list[list[str]]: return [(re.split('(?<=.)(?=[A-Z])', x) if isinstance(x, str) else x) for x in seq] def _split_peptide(peptide: (str | list[str])) -> list[str]: if (not isinstance(peptide, str)): return peptide return re.split('(?<=.)(?=[A-Z])', peptide) def matches_precursor(self, seq: (str | list[str]), prec_mass: float, prec_charge: int, prec_tol: int=50) -> tuple[(bool, list[float])]: seq_mass = self._mass(seq, charge=prec_charge) delta_mass_ppm = [self._calc_mass_error(seq_mass, prec_mass, prec_charge, isotope) for isotope in range(self.isotope_error_range[0], (self.isotope_error_range[1] + 1))] return (any(((abs(d) < prec_tol) for d in delta_mass_ppm)), delta_mass_ppm) def compute_aa_er(self, peptides_truth: (list[str] | list[list[str]]), peptides_predicted: (list[str] | list[list[str]])) -> float: peptides_truth = self._split_sequences(peptides_truth) peptides_predicted = self._split_sequences(peptides_predicted) return float(jiwer.wer([' '.join(x) for x in peptides_truth], [' '.join(x) for x in peptides_predicted])) def compute_precision_recall(self, targets: (list[str] | list[list[str]]), predictions: (list[str] | list[list[str]]), confidence: (list[float] | None)=None, threshold: (float | None)=None) -> tuple[(float, float, float, float)]: targets = self._split_sequences(targets) predictions = self._split_sequences(predictions) (n_targ_aa, n_pred_aa, n_match_aa) = (0, 0, 0) (n_pred_pep, n_match_pep) = (0, 0) if ((confidence is None) or (threshold is None)): threshold = 0 confidence = np.ones(len(predictions)) for i in range(len(targets)): targ = self._split_peptide(targets[i]) pred = self._split_peptide(predictions[i]) conf = confidence[i] if (pred[0] == ''): pred = [] n_targ_aa += len(targ) if ((conf >= threshold) and (len(pred) > 0)): n_pred_aa += len(pred) n_pred_pep += 1 n_match = self._novor_match(targ, pred) n_match_aa += n_match if ((len(pred) == len(targ)) and (len(targ) == n_match)): n_match_pep += 1 pep_recall = (n_match_pep / len(targets)) aa_recall = (n_match_aa / n_targ_aa) if (n_pred_pep == 0): pep_precision = 1.0 aa_prec = 1.0 else: pep_precision = (n_match_pep / n_pred_pep) aa_prec = (n_match_aa / n_pred_aa) return (aa_prec, aa_recall, pep_recall, pep_precision) def calc_auc(self, targs: (list[str] | list[list[str]]), preds: (list[str] | list[list[str]]), conf: list[float]) -> float: (x, y) = self._get_pr_curve(targs, preds, conf) (recall, precision) = (np.array(x)[::(- 1)], np.array(y)[::(- 1)]) width = (recall[1:] - recall[:(- 1)]) height = np.minimum(precision[1:], precision[:(- 1)]) top = np.maximum(precision[1:], precision[:(- 1)]) side = (top - height) return ((width * height).sum() + (0.5 * (side * width).sum())) def _get_pr_curve(self, targs: (list[str] | list[list[str]]), preds: (list[str] | list[list[str]]), conf: np.ndarray, N: int=20) -> tuple[(list[float], list[float])]: (x, y) = ([], []) t_idx = np.argsort(np.array(conf)) t_idx = t_idx[(~ conf[t_idx].isna())] t_idx = (list(t_idx[((t_idx.shape[0] * np.arange(N)) / N).astype(int)]) + [t_idx[(- 1)]]) for t in conf[t_idx]: (_, _, recall, precision) = self.compute_precision_recall(targs, preds, conf, t) x.append(recall) y.append(precision) return (x, y) def _mass(self, seq: (str | list[str]), charge: (int | None)=None) -> float: seq = self._split_peptide(seq) calc_mass = (sum([self.residues[aa] for aa in seq]) + H2O_MASS) if (charge is not None): calc_mass = ((calc_mass / charge) + PROTON_MASS_AMU) return calc_mass def _calc_mass_error(self, mz_theoretical: float, mz_measured: float, charge: int, isotope: int=0) -> float: return (((mz_theoretical - (mz_measured - ((isotope * CARBON_MASS_DELTA) / charge))) / mz_measured) * (10 ** 6)) def _novor_match(self, a: list[str], b: list[str]) -> int: n = 0 mass_a: list[float] = [self.residues[x] for x in a] mass_b: list[float] = [self.residues[x] for x in b] cum_mass_a = np.cumsum(mass_a) cum_mass_b = np.cumsum(mass_b) (i, j) = (0, 0) while ((i < len(a)) and (j < len(b))): if (abs((cum_mass_a[i] - cum_mass_b[j])) < self.cum_mass_threshold): n += int((abs((mass_a[i] - mass_b[j])) < self.ind_mass_threshold)) i += 1 j += 1 elif (cum_mass_a[i] > cum_mass_b[j]): i += 1 else: j += 1 return n
def _dataset_exists(path, annotation, image_dir): if (not osp.exists(path)): logger.debug('Config dataset_dir {} is not exits, dataset config is not valid'.format(path)) return False if annotation: annotation_path = osp.join(path, annotation) if (not osp.isfile(annotation_path)): logger.debug('Config annotation {} is not a file, dataset config is not valid'.format(annotation_path)) return False if image_dir: image_path = osp.join(path, image_dir) if (not osp.isdir(image_path)): logger.warning('Config image_dir {} is not a directory, dataset config is not valid'.format(image_path)) return False return True
def create_optimizer(loss, init_lr, num_train_steps, num_warmup_steps, use_tpu): global_step = tf.train.get_or_create_global_step() learning_rate = tf.constant(value=init_lr, shape=[], dtype=tf.float32) learning_rate = tf.train.polynomial_decay(learning_rate, global_step, num_train_steps, end_learning_rate=0.0, power=1.0, cycle=False) if num_warmup_steps: global_steps_int = tf.cast(global_step, tf.int32) warmup_steps_int = tf.constant(num_warmup_steps, dtype=tf.int32) global_steps_float = tf.cast(global_steps_int, tf.float32) warmup_steps_float = tf.cast(warmup_steps_int, tf.float32) warmup_percent_done = (global_steps_float / warmup_steps_float) warmup_learning_rate = (init_lr * warmup_percent_done) is_warmup = tf.cast((global_steps_int < warmup_steps_int), tf.float32) learning_rate = (((1.0 - is_warmup) * learning_rate) + (is_warmup * warmup_learning_rate)) optimizer = AdamWeightDecayOptimizer(learning_rate=learning_rate, weight_decay_rate=0.01, beta_1=0.9, beta_2=0.999, epsilon=1e-06, exclude_from_weight_decay=['LayerNorm', 'layer_norm', 'bias']) if use_tpu: optimizer = tf.contrib.tpu.CrossShardOptimizer(optimizer) tvars = tf.trainable_variables() grads = tf.gradients(loss, tvars) (grads, _) = tf.clip_by_global_norm(grads, clip_norm=1.0) train_op = optimizer.apply_gradients(zip(grads, tvars), global_step=global_step) new_global_step = (global_step + 1) train_op = tf.group(train_op, [global_step.assign(new_global_step)]) return (train_op, learning_rate)